••
-
••
-
THE MASSACHUSETTS
TOXICS USE REDUCTION INSTITUTE
EVALUATION AND IMPLEMENTATION
OF NO-CLEAN SOLDER PASTE FOR
SURFACE MOUNT TECHNOLOGY
WITH
A Focus ON SMALL CONTRACT
MANUFACTURERS
GRADUATE THESIS
Technical Report No. 40
1997
University of Massachusetts Lowell
Evaluation and Implementation of No-CleaQ Solder for
Surface Mount Technology
With a Focus on Small Contract Manufacturers
Doug Sommer
Prof. Sammy G. Shina
Mechanical Engineering Dept.
University of Massachusetts Lowell
The 1996 - 1997 Toxics Use Reduction Research Fellows Program
The Toxics Use Reduction Institute
University of Massachusetts Lowell
1997
••
TOXICS USE
REDUCTION
INSTITUTE
••
- -
All rights to this report belong to the Toxics Use Reduction Institute. The
material may be duplicated with permission by contacting the Institute.
The Toxics Use Reduction Institute is a multi-disciplinary research, education,
and policy center established by the Massachusetts Toxics Use Reduction Act of
1989. The Institute sponsors and conducts research, organizes education and
training programs, and provides technical support to promote the reduction in
the use of toxic chemicals or the generation of toxic chemical byproducts in
industry and commerce. Further information can be obtained by writing the
Toxics Use Reduction Institute, University of Massachusetts Lowell, One
University Avenue, Lowell, Massachusetts 01854.
°Toxics Use Reduction Institute, University of Massachusetts Lowell
Toxics Use Reduction Institute
Research Fellows Program
In 1991, the Toxics Use Reduction Institute established the Research Fellows Program at the
University of Massachusetts Lowell (UML). The Research Fellows Program funds toxics use
reduction projects performed by a graduate student and his/her advisor. The goals of the
Research Fellows Program are:
•
•
•
•
•
to develop technologies, materials, processes, and methods for implementing toxics use
reduction techniques
to develop an understanding of toxics use reduction among UML graduate students and
faculty
to facilitate the integration of the concept of toxies use reduction into UML research
projects
to provide UML faculty with "incubator" funding for toxies use reduction related
research, and
to act as a liaison between Massachusetts industries and UML faculty.
Notice
This report has been reviewed by the Institute and approved for publication. Approval does
not signify that the contents necessarily reflect the views and policies of the Toxics Use
Reduction Institute, nor does the mention of trade names or commercial products constitute
endorsement or recommendation for use.
ABSTRACT
For many decades, the electronics industry has utilized ozone depleting solvents to clean
electronics assemblies. The cleaning of electronic assemblies has been necessary to ensure the
removal of performance detrimental contaminants, such as soldering residues, that reduce the
reliability of the electronic assembly. As the usage of these cleaning solvents continued, the
harmful effects on the environment have become increasingly apparent. Legislation has forced
electronics manufacturers to consider and adopt new alternatives from the environmentaJly
deleterious cleaning practices previously deemed acceptable. One of the most advantageous
methods to accomplish this has been to adopt a no-clean approach to soldering.
As the name implies, electronics assemblies that are soldered using no-clean materials do
not need to be cleaned after assembly operations. The residues that they leave behind are benign
to the end use reliability of the assembly. They also have inherent other benefits such as the
elimination of cleaning equipment, waste effluent, and all its other related costs. Unfortunately,
implementation of no-clean soldering has proven to be difficult for smaller assembly shops due
to the many concerns associated with no-clean soldering. Many of these manufacturers lack the
engineering resources, economic means, and the time to transition to a new process without loss
of quality, throughput, and cost.
This study addresses the implementation of no-elean solder pastes for a small contract
manufacturing situation. The issues regarding the cleaning of electronics are discussed as well
as the mechanics of the no-clean materials involved. A general methodology for implementation
was generated along with a research implementation methodology. Following the research
implementation methodology, materials were qualified against industry standards for
acceptability. Matrix experiment methods were utilized to model the manufacturing equipment
at an actual small contract assembly house in Massachusetts. Results led to the statistical
examination of their equipment and yielded the important factors involved with each piece of
equipment. Optimization techniques were utilized to recommend the best settings for the
manufacturing equipment, and the settings found were tested on a real world product of the
contract assembler. The results obtained were excellent and the customer has subsequently
changed the requirements of that particular product to be assembled exclusively using no-clean
solder pastes. An economic analysis was also completed to determine the cost savings that
would be realized if conversion was completed at the contract assembly house.
11
ACKNOWLEDGMENTS
There are many individuals who had input in to this project without whom would have made it
difficult or impossible to complete.
Extra special thanks to my brother Thomas for his technical and personal guidance throughout
this project. Many aspects of this project would never have worked if it wasn't for your constant
help and guidance.
Special thanks to Professor Sammy Shina for being my faculty advisor during this study. Your
technical guidance was invaluable for completing this project.
Special thanks also to Monica Becker and the Toxic Use Reduction Institute for sponsoring this
project. This project would never have been possible without your financial assistance and
understanding. I have a new appreciation for the environmental issues which industry faces
thanks to your guidance.
Thanks to Tristar Technologies Incorporated for the opportunity to use your equipment, perform
the study in your facility, and the donation of test circuit boards. Your engineering support; Bob
Laurenti and Lisa Labbe, were invaluable to completing the experiments. Lisa- Thanks for the
materials.
Thanks also to the many other industry engineers, and suppliers whose donations of materials
and technical advise made this study possible.
Thanks to Les Hymes, for your wonderful seminar and donation of IPC stadnards.
Thanks to Tim O'neal of ESP solders for your "value added" services and technical advise.
Thanks to Greg Tashjian of AT&T (Now Lucent Technology) for taking the time to steer the
project in the right direction.
Thanks to Robby Zylich of Bull Information Systems for the use of your software
Thanks to Dr. Brett E. Hardie for your technical expertise.
Thanks to Les Hymes for the generous donation of IPC acceptability standards.
And finally, thanks to Doug Peck of AEIC for your technical support and breakfast at Bickfords.
111
TABLE OF CONTENTS
Disclaimer
i
Abstract
ii
Acknowledgments
iii
List of Figures
ix
List of Tables
xi
List of Abbreviations
xiii
Chapter 1: Introduction
1.1
Background
1
1.2
Background on Soldering
2
1.3
Background on Surface Mount Technology
2
1.4
Surface Mount Technology Assembly Process
4
Chapter 2: Cleaning Electronic Assemblies
2.1
Reasons for Cleaning Electronic Assemblies
8
2.2
Sources of Contaminants
8
2.3
Environmental Impacts From Cleaning Electronics
9
2.4
Current Cleaning Options Available
11
Chapter 3: Solder Paste Mechanics
3.1
Solder Paste Background
12
3.2
Solder Alloy
13
IV
14
3.2.1 Solder Alloy Particle Size
3.3
14
Flux Binder
3.3.1 Flux Binder Solids Portion
14.
3.3.2 Flux Binder Solvents and Modifiers Portion
15
3.3.3 Flux Binder Activators Portion
15
Chapter 4: No-Clean Soldering
4.1
No-Clean Background
17
4.2
Flux and No-Clean Flux Materials
18
4.3
No-Clean Advantages
19
4.4
No-Clean Concerns
20
Chapter 5: Implementation Methodology
5.1
Introduction
22
5.2
General Implementation Strategy
22
5.3
Research Implementation Methodology
26
Chapter 6: Material Qualification ad Evaluation Results
6.1
Introduction
29
6.2
Pre-Reflow Evaluation Results
31
6.3
6.2.1 Appearance
31
6.2.2 Tackiness and Adhesion
31
6.2.3 Slump
33
6.2.4 Dispensability
34
Post Reflow Evaluation Measures
34
6.3.1 Wettability of Flux (Flux Efficacy)
34
6.3.2 Solder Balling
35
6.3.3 Electrical Cleanliness (Ionic Contamination)
38
6.3.3.1 Resistivity of Solvent Extract
38
6.3.3.2 Surface Insulation Resistance
40
6.3.3.2.1
SIR Theory
41
6.3.3.2.2
Experimental Setup
42
v
6.3.3.2.3
6.4
6.5
Experimental Results
43
43
Significant Tests Not Performed
6.4.1 Electro-Chemical-Migration
44
6.4.2 Pre-Reflow Corrosion
44
6.4.3 Halide Presence
44
45
Paste Evaluation Conclusions
Chapter 7: Stencil Printing Analysis
7.1
Introduction
47
7.2
Stencil Printer Equipment
47
7.3
Stencil Printer Parameter Selection for Analysis
48
7.4
Designed Experiment Analysis and Results
52
7.5
7.4.1 Average Analysis
53
7.4.2 Variability Analysis
55
Stencil Printer Analysis Conclusions
59
Chapter 8: Full SMT Process Analysis
8.1
Introduction and Approach
60
8.2
Qualification Metric and Layout of Matrix Array
60
8.3
Experimental Parameters
61
8.3.1 Stencil Printer Parameters
61
8.3.2 Pick and Place Parameters
62
8.3.3 Reflow Oven Parameters and Other Factors
63
8.4
Standard Pitch Results and Discussion
65
8.5
Fine Pitch Analysis and Discussion
67
8.6
Ultra-Fine Pitch Results and Discussion
70
8.7
Full Process DOE Conclusions
70
8.8
Confirming Experiments Results and Discussion
70
8.9
Implementation Conclusions and Recommendations
72
VI
8.10
Future Implementation at Tristar Technologies
73
Chapter 9: Financial Implications of Conversion
9.1
Introduction
74
9.2
Financial Impacts at Tristar Technologies
74
9.3
Aqueous Cleaner Related Costs
74
9.3.1 Electricity Consumption
76
9.3.2 Water Consumption
77
9.3.3 Processing Labor During Cleaning
78
9.3.4 Filter Replacement
79
9.3.5 Miscellaneous Maintenance Materials
80
9.3.6 Maintenance Labor Costs
80
9.3.7 Floor Space
81
9.3.8 Cleaner Equipment Capital Cost
81
9.3.9 Aqueous Cleaner Total Costs
82
9.3.10 Cleaner Operations Savings From Percentage
82
ConversioI\ to No-Clean
9.4
No-Clean Soldering Material Conversion Costs
·83
9.5
Total Cost Savings or Liability for No-Clean Conversion
84
9.6
Financial Implications of Conversion Conclusions
85
Chapter 10: Overall Conclusions and Recommendations
10.1
Overall Conclusions
86
10.2 Recommendations for Future Research
86
Appendix A
I-A
Appendix B: IPC Industry Standard Test Methods
I-B
IPC-TM-650-2.4.35 Solder Paste Slump Test
Vll
1-B
IPC-TM-650-2.4.45 Solder Paste Wetting Test
3-B
IPC-TM-650-2.6.15 Corrosion, Flux
4-B
IPC-TM-650-2.4.43 Solder Paste Solder Ball Test
6-B
IPC-TM-650-2.3.25 Resistivity of Solvent Extract
9-B
IPC-TM-650-2.6.3.3 Surface Insulation Resistance, Fluxes
11-B
Appendix C: List of Industry Contacts
VIlI
1-C
LIST OF FIGURES
Figure 1: Comparison Between Equivalent SMT
and Through Hole Integrated Circuits
3
Figure 2: Typical SMT Process Flow
5
Figure 3: Typical Convection Reflow Profile
6
Figure 4: Solder Paste Constituents
12
Figure 5: Tin Lead Phase Diagram
13
Figure 6: General Implementation Approach
23
Figure 7: Research Implementation Methodology
28
Figure 8: Tristar Donated Ultra-Fine Pitch SMT Test Board
30
Figure 9: Topline Brand SMT Test Board
30
Figure 10: Paste"A" Dewetting Example
35
Figure 11: Graph of Solder Ball Test Results
36
Figure 12: Visual Comparison Between Solder Balling Results
37
Figure 13: Alpha Brand Omega Meter
39
Figure 14: Adjacent Conductors For SIR Testing
41
Figure 15: Military "Y" Coupon Used for SIR Testing
42
Figure 16: Experimental SIR Set Up
43
Figure 17: Dek 249 Semi-Automatic Stencil Printer
48
Figure 18: Cause and Effect Diagram of Potential Factors Affecting Stencil Printing
49
Figure 19: Test Pattern Area on Topline Test Boards for Stencil Printer Analysis
51
Figure 20: Linear Graph of Factors for Column Assignment in L8 Array
51
Figure 21: Plot of Primary Factor Average Effects
54
Figure 22: Plot of Primary Factor Variability Effect
57
Figure 23: Comparison Between Parameter Contribution
Effecting Average and Variability
59
Figure 24: Rating Criterion for Solder Joint Fillet Geometry
61
Figure 25: Mydata TP9-UFP Pick and Place Machine
63
Figure 26: Heller 1500 Forced Convection Reflow Oven
63
Figure 27: Assignment of Parameters to Array for Full Process Analysis
64
IX
Figure 28: Test Site Used for Standard Pitch Measurement on Topline Test Boards
65
Figure 29: Cross-Section of Standard Pitch Lead
66
Figure 30: Test Site Used for Fine Pitch Measurement on Topline Test Boards
67
Figure 31: Cross-Section of Fine Pitch Component Lead
68
Figure 32: Real World PCB Used for Confirming Experiment
71
Figure 33: Assembled Real World Test Circuit Board
72
Figure 34: Trieber TRM-SMD Series 400 In-Line Cleaner
75
Figure 35: Trieber water Recycling System
75
Figure A-I: Interactionn Effect on Average For Dqueegee Pressure and Speed
I-A
Figure A-2: Interaction Effect on Average For Squeegee Pressure and Material
2-A
Figure A-3: Interaction Effect on Variablity For Squeegee Pressure and Speed
2-A
Figure A-4: Interaction Effect on variability For Squeegee Pressure and Material
3-A
x
LIST OF TABLES
Table 1: Classification of Chief Contaminates on Electronic Assemblies
9
Table 2: Current Electronics Cleaning Alternatives
II
Table 3: Classification of Soldering Materials By Main Ingredients
18
Table 4: Typical No-Clean Specification Breakdown
19
Table 5: Subjective Rating for Paste Tackiness Ability
33
Table 6: Experimental Solvent Extract Resistivity Results
40
Table 7: Paste Evaluation Rating Matrix
45
Table 8: Selected Parameters and Levels
50
Table 9: L8 Array with Chosen Parameters and Levels
52
Table 10: Experimental Results
53
Table 11: Average effects of Primary Factors
53
Table 12: ANOVA Results for Average Data
55
Table 13: Percent Contribution of Factors Affecting Average
55
Table 14: Calculated SIN Values for Results
56
Table 15: Variability Effect of Primary Factors
57
Table 16: ANOVA Results for SIN Variability Data·
58
Table 17: Percent Contribution of Factors Effecting Variability
58
Table 18: Selected Parameters and Levels for Full Process DOE Analysis
65
Table 19: Results for Standard Pitch DOE
66
Table 20: Experimental Results for Fine-Pitch DOE
68
Table 21: ANOVA Results for Fine Pitch Process DOE
69
Table 22: Tristar Cleaner System Related Cost Breakdown
76
Table 23: Aqueous Cleaner Electricity Consumption Per Hour
77
Table 24: Water Consumption Cost Calculation
78
Table 25: Filter Replacement Cost Breakdown Per Work Hour Used
79
Table 26: Miscellaneous Cleaner Material Maintenance Cost
80
Table 27: Total Cleaner operation Cost Per Work Hour
82
Table A-I: Solder Balling Test Results
I-A
Table A-2: Interaction Effect on Average For Squeegee Pressure and Speed
I-A
Xl
Table A-3: Interaction Effect on Average For Squeegee Pressure and Material
I-A
Table A-4: Interaction Effect on Variability For Squeegee Pressure and Speed
2-A
Table A-5: Interaction Effect on Variability For Squeegee Pressure and Material
3-A
Xll
List of Abbreviations
ANOVA:
ANalysis Of VARiance; A statistical approach to analyzing the variability and
.
percent contributions of factors in a matrix experiment.
CFC:
Chloro-Fluoro-Carbon; CFCs are compounds consisting of chlorine, fluorine,
and carbon, which are very stable in the atmosphere near ground level because
they do not readily break down under the action of the sun. They were utilized in
chemical solvents commonly up until the early 1990's due to their excellent
stability and cleaning properties. These compounds eventually ascend up to the
stratospheric ozone layer and react detrimentally with the protective ozone
compound and cause ozone depletion. It has subsequently been banned from
production in many countries including the United States.
DOE:
Design Of Experiments; A set of experiments in which the settings of various
process parameters are changed and the corresponding results obtained are
analyzed statistically. The systematic approach inherently allows for parameter
contribution analysis, interaction analysis, and process optimization.
FP-SMT:
Fine Pitch Surface Mount Technology: The usage of surface mounted electronic
components which have a lead pitch spacing center to center below 50 mils and
greater than 20 mils.
HCFC:
Hydro-Chloro-Fluoro-Carbon; HCFCs are compounds that are composed of
chlorine, fluorine, carbon, and hydrogen. They are generally less stable than CFC
based compounds, and the majority of them break down before they can ascend
up to the stratospheric ozone layer. HCFCs typically have an ozone depletion
potential of between 2% to 10% ofCFCs.
IPC:
The Institute for Packaging and interConnection; The IPC is the group
responsible for generating the most widely adopted acceptability standards for the
electronics industry.
ODC:
Ozone Depleting Compound; Any chlorine and carbon containing compound
such as CFC, HCFC, or others that has the potential to ascend to the stratospheric
ozone layer and react adversely with the ozone compound and cause ozone
depletion.
ODP:
Ozone Depletion Potential; ODP is the relative value ofa substance's potential
to deplete stratospheric ozone. It is compared to the reference gas
trichlorofluoromethane (CFCll), which has an ODP of one (1).
Xlll
List of Abbreviations Continued
PCB:
Printed Circuit Board~ A rigid substrate on which components are placed or
inserted that has electrical traces on or within the substrate to allow for electrical
connection between components.
RMA:
Rosin Mildly Activated~ A common flux used for electronics assembly which
uses colophony (rosin), a distillate of pine trees, as the major solids portion
ingredient. Its inherent benign residue properties are what has led to research in
to no·clean fluxes.
SMT:
Surface Mount Technology; An assembly technology in which components are
placed on to electrically conductive areas on the surface of the printed circuit
board instead of through plated through holes in the substrate. SMT is used
primarily for its advantages over through hole technology for increased
automation facilitation and increased circuit densities possible.
UFP-SMT:
Ultra Fine Pitch Surface Mount Technology~ The usage of surface mounted
electronic components which have a lead pitch spacing center to center of 20 mils
or below.
VOC:
Volatile Organic Compound~ VOCs contain one or more carbon atoms, and they
may also contain chlorine, oxygen, and nitrogen atoms. They exist as gases or
volatile liquids and are not as stable as CFCs, allowing them to break down at
lower levels in the atmosphere than CFCs. Many common compounds such as
isopropyl alcohol are considered VOCs.
WS, OA:
Water Soluble, Organically Acid~ A classification of fluxes which utilize
organic acids as activators and are generally water soluble or soluble with water
and a soponifier added.
XIV
CHAPTER 1: INTRODUCTION
1.1 BACKGROUND
The electronics industry has changed and grown significantly in the pastfew
decades. Along with great technological advancements however, growing environmental
concerns over the cleaning of electronics have become an increasingly troublesome issue.
Cleaning electronic assemblies is typically accomplished to ensure removal of
performance detrimental contaminants such as printed circuit board fabrication and
assembly soldering residues. Historically, electronics manufacturers have utilized ozone
depleting chlorinated solvents for cleaning after assembly operations. As the usage of
these chlorinated solvents continued over the years, the harmful effects on the
environment of these chemicals have become apparent. Recently, many federal and
international regulations, such as the Montreal Protocol, have emerged to restrict or
eliminate the use of such chemicals.
A large part of the Massachusetts industry is centered around electronics
manufacturing, and these environmental concerns affect them greatly. Many of the larger
companies have already changed to more environmentally friendly manufacturing
techniques. However, the changeover for some of the smaller companies to more
environmentally friendly manufacturing techniques has shown to be very difficult. Many
of these companies lack the engineering resources, economic means, and the time to
transition to a new process without loss of quality, throughput, and cost.
One solution towards eliminating chlorinated solvent usage has been to adopt a
no-clean soldering process. Here the idea is to use soldering materials which leave
behind residues that do not compromise the end reliability of the assembly, and to leave
them on the assembled printed circuit board. Thus, the cleaning step is completely
eliminated, along with the environmentally damaging cleaning solvents and other related
waste effluents. No-clean is the best long term approach to the problem and has other
inherent advantages such as eliminating the capital investment of cleaning equipment as
well as all its associated operating costs. Unfortunately, there are many concerns and
1
challenges to implementing a no-clean process because of tighter process control
requirements, end use reliability concerns, and material uncertainties.
A great deal of information is available concerning no-clean, although it is
difficult to understand how the material and implementation issues relate to each
particular manufacturing setting. A consolidation of information about no-clean and a
technique for implementation at a small electronic contract manufacturing company
would be a valuable tool for other Massachusetts companies interested in switching to a
no-clean process. The subject of this study is the qualification, evaluation, and
implementation methods used to successfully implement a no-clean surface mount
process at a local, low volume, high mix contract manufacturing company.
1.2 BACKGROUND ON SOLDERING
Soldering is one of the oldest methods ofjoining metals. Soldering utilizes a
filler metal to join two or more base metals which have a melting point above the filler
metal. During soldering, a chemical called a flux, is applied to the base and filler metals
which removes oxidation from them at elevated temperatures. This cleaning (fluxing)
action is a required property for soldering. Once the solderable surfaces have been
cleaned through this fluxing action, the melted solder alloy can then bond to the base
metals resulting in an intermetallic bond. This intermetallic bond is a thin layer in which
the solder alloy and the basis metals alloy to themselves and form an adhesive bond. In
electronics, the solder alloy is utilized for its property of having both good electrical
conductivity and good mechanical bonding strength to hold electronic components in
place.
1.3 BACKGROUND ON SURFACE MOUNT TECHNOLOGY
As the name implies, surface mount technology (SMT) is an assembly technology
in which electronic components are placed on to electrically conductive areas on the
surface of the printed circuit board instead of through plated through holes in the
substrate (through hole technology). This concept has been utilized in hybrid assembly
since the 1960's by interconnecting chip resistors, chip capacitors, and bare
2
semiconductor dies on hybrid substrates. However, the potential of surface mount
technology was not fully utilized and explored until the early 1980's. It has shown to be
one of the most significant developments in the electronics era. I
Surface mount technology has many inherent advantages over through hole
technology.
Some of the merits of an electronic assembly utilizing surface mount
technology as compared with an equivalent through hole assembly include:
- Reduced weight,
- Reduced size,
-Increased circuit density,
-Improved automation facilitation,
.-Improved electrical performance,
-Lower costs in volume production2
In an SMT assembly, the majority of component interconnection traces on the PCB are
routed on intermediate layers of the circuit board which allow much greater real estate
efficiency and permit usage of both sides of the board for component population. Factors
such as shorter leads, and lead sizes required allow for component miniaturization to a
degree which is simply not possible with through hole components. An example of this
can be seen in Figure 1, which compares a through hole integrated circuit to its SMT
equivalent. Note the size difference between the two packages.
FIGURE 1: Comparison Between Equivalent SMT and Through Hole Integrated Circuits 3
Hwang, Jennie. Solder Paste in Electronics Packaging. Van Nostrand Reinhold, New York, 1992. Page 9.
Hwang, Page 9.
3 Prasad, Ray P. Surface MOUflt Technology. Van Nostrand Reinhold, New York, 1989. Page 6.
I
2
3
While through hole technology will continue to maintain usage within a large
segment of the industry for various reasons, this brief presentation illustrates the simple
point that SMT is where the market continues to move and where study should be
focused. Even more specifically, fine pitch (FP-SMT) and ultra-fine pitch surface mount
technology (UFP-SMT) is where study should be focused. FP-SMT is defined as the
usage of lead pitch spacing in electronic components that is below 50 mils but above 20
mils, while UFP-SMT is defined as the usage of lead pitch spacing in electronic
components that is below 20 mils. Presently, FP-SMT and UFP-SMT represents the
widest used technological solution to the ubiquitous problem of miniaturization versus
the pervasive push for increased input/output connections for integrated circuit packages.
1.4 SURFACE MOUNT TECHNOLOGY ASSEMBLY PROCESS
At its most basic level, the typical surface mount assembly process follows a
series of fundamental discreet steps. Controlled volumes of solder in a paste form are
placed on to electrically conductive land areas of the circuit board. Components are then
placed on to the deposited solder paste which has sufficient tackiness to "hold" them in
place. The entire assembly is then subjected to a thermal cycle which is called reflow.
During the reflow process, the solder paste melts and coalesces to "wet" both the
component leads and the printed circuit land areas and ultimately form the mechanical
and electrical solder joint between the two. Traditionally, the final step has been to
remove soldering residues from fluxes that are left behind after reflow by cleaning the
assembly. Inspection, touch up, electrical test, or other operations may follow, but are
not generally considered part of the SMT assembly process.
The process flow previously described is purposely generic. It does not specify
how the paste is applied, through which process the components are placed, which
technique for reflow is used, or how the assembly is cleaned. As SMT has evolved, many
methods have been developed to suit the many diverse assembly types and manufacturing
settings which exist. While there are many different techniques available for each step of
4
assembly, the one which will be studied here conforms to the event flow shown in Figure
2. This process flow is the most commonly used for SMT assembly in the industry.
I
Stencil Print
I
Place
Solder Paste -~ Components
Convection
IMass Reflow
I
I--~
I
~i
"I
II----I~
Clean
Assembly
FIGURE 2: Typical SMT Process Flow
During the stencil printing process, a metal stencil which has openings called
apertures is placed over the circuit board. The apertures on the stencil are aligned
corresponding to areas on the circuit board where solder paste is to be deposited. The
solder paste is then forced through the apertures by a squeegee as it wipes across the
stencil. Finally, the stencil is removed to leave the deposited paste "bricks" on the circuit
board. This method of deposition works very well once the printer is set correctly and
proper process control guidelines are followed. Since printing is the foundation for all
stages to follow, obtaining a satisfactory print is essential.
The next process step is component placement on to the solder bricks that were
previously deposited. The paste has the property of being tacky, which holds the
components in place once situated. Components can be placed either manually, semimanually, or by a machine. Generally automation is accepted as the preferable alternative
for re,asons of repeatability, accuracy, and speed. Many types of machinery have been
developed to quickly and accurately pick a component from feeders, orient it, and place it
precisely on to the paste. Obviously, budget, product mix, volumes, setup times,
handling difficulty, and other factors influence the utility of a pick and place machine.
The next step in the process is convection mass reflow. Convection heat transfer
has become the most widely accepted method for reflowing solder paste. Other methods
for reflow include infrared, laser, and vapor phase reflow. A typical convection oven has
multiple upper and lower zones which emit heat at set temperatures. The assembly to be
reflowed is placed on a moving belt or rail which moves it through the oven at a constant
5
speed. As the assembly passes through each zone, it is subjected to a specific thermal
profile which serves many purposes for soldering. A generic convection reflow profile
can be seen in Figure 3.
t
Zone 1:
Zone 2:
Pre-Heat
Soak
Zone 3:
Reflow
Zone 4:
Cooldown
T
E
M
P
TIME~
FIGURE 3: Typical Convection Reflow Profile
In the first zone, called the preheat zone, the temperature of the assembly is slowly
ramped upwards in order to not thermally shock the board or components which could
lead to unwanted stress and mechanical failure. The next zone is called the soak zone,
which serves the function of evaporating solvents from the solder paste and allowing the
flux in the paste to remove oxidation from the surfaces to be soldered. As the name
implies, during the soak zone, the temperature change is kept fairly constant. After soak,
the next zone is reflow. During reflow, the temperature is raised above the melting point
of the solder alloy for a set period of time while the solder becomes liquidous and flows
to generate the solder joint. The final zone is cool down, in which the temperature of the
assembly is lowed back to room temperature at a relatively slow rate as to not thermally
shock the board and components while cooling.
6
To properly solder, every paste manufactured is designed to have a set reflow profile
which it should be subjected to. Oven zone temperatures must be programmed and
changed for boards with large differences in component densities. Larger or increased
component quantities on an assembly absorb more heat and thus increase the amount of
heat that is needed for the assembly to realize the same thermal profile as it would with
smaller or less components.
Traditionally, cleaning has been the final step in the assembly process. The
cleaning of electronic assemblies has been necessary in the past to remove potentially
harmful contaminants from the product. A detailed discussion of cleaning electronic
assemblies is covered in Chapter 2.
7
CHAPTER 2: CLEANING ELECTRONIC ASSEMBLIES
2.1 REASONS FOR CLEANING ELECTRONIC ASSEMBLIES
Cleaning electronic assemblies has been necessary for a variety of reasons. The
principle motive for cleaning is to remove potentially hazardous contaminants from the
assembly which could adversely affect the reliability of the product. Depending on the
type of contamination present, electrical degradation, corrosion, or mechanical
damage/interference may occur over the life span of the device. The simplest approach to
significantly reducing failure caused by these sources has simply been to tolerate a
cleaning operation.
2.2 SOURCES OF CONTAMINANTS
The sources of contaminants found on printed wiring assemblies can come from
many sources and can be broken down in to many categories. Chemically, these
categories can be classified as particulate, ionic, nonionic, polar, nonpolar, organic or
inorganic. 4 Table 1 shows the classification of the chief types of contaminates found on
both bare printed circuit boards (PCB's) and final assemblies.
4
Hymes, Les. Cleaning Printed Wiring Assemblies in Today's Environment. New York, 1991. Page 78.
8
TABLE 1: Classification of Chief Contaminants on Electronic Assemblies 5
Flux activators
Flux resin / rosin
Activator residues
Grease
Dust
Soldering salts
Surfactants (non-ionic)
Handling soils
Handling soils
Synthetic polymers
Lint
Residual plating salts
Soldering oils
Hair/skin
Surfactants (ionic)
Metal oxides
Neutralisers
Handling soils
Ethanolomines
Hand creams
Resin and fiberglass debris from board
fabrication
Metal/plastic chips from machining
operations
Silicones
Ionic species are those that in solution form charged species and thus conduct electricity.
A polar molecule is by definition one having a dipole moment. This means that there
exists an uneven distribution of electrons in the molecule which produces a dipole
molecule. 6
Depending on the end use of the electronics assembly, every contaminate has the
potential to be detrimental to product reliability. Contaminants can be damaging
mechanically, electrically, and chemically. Obviously, the higher up on the quality
spectrum a product lies, such as a military or medical product, the more imperative the
quality and reliability becomes.
2.3 ENVIRONMENTAL IMPACTS FROM CLEANING ELECTRONICS
Traditionally, electronic assemblies have been cleaned with solvents that contain
ozone depleting compounds (ODCs). ODCs are compounds which contain chlorine and
carbon, which are very stable in the environment near ground level because they do not
5
6
Hymes, Page 79.
Hymes, Page 79.
9
readily break down under the action of the sqn. It was common place to see the usage of
chemicals such as CFC-113, and 1,1,1 Trichloroethane in electronic assembly houses for
many years during the 1980's and early 1990's. They were used on a widespread basis
due to favorable inherent properties such as excellent stability, material compatibility,
relatively low toxicity, non-flammability, and most importantly superb cleaning ability.
As these solvents would evaporate, they would eventually ascend to the ozone layer in
the upper stratosphere because they were so stable near ground level they would not
break down beforehand. Once these ODCs would reach the stratospheric ozone layer,
they would react detrimentally with the ozone molecule and cause ozone depletion.
Chloroflouorocarbons (CFCs), and hydrochlorofluorocarbons (HCFCs) were the major
classifications of ODCs which were utilized in solvents. Depending upon the chemical
makeup ofthe solvent, each ODC had a varying degree of detrimental effect upon the
ozone molecule. This varying detrimental effect is usually expressed in terms of a
solvents ozone depletion potential (ODP), as compared to the reference gas
trichlorofluoromethane (CFC11), which has an ODP of one. CFCs are generally
considered to have the highest ODP numbers, while HCFCs have ODPs of typically 2%
to 10% ofCFCs. 7
As the usage of these ODCs has continued over the years, the harmful effects on
the ozone layer in the upper stratosphere have become very apparent. As the ozone layer
is slowly depleted, it allows more ultraviolet radiation from the sun to penetrate through
the atmosphere and ultimately cause global warming. 8
Recent years have seen the introduction and adoption of many federal and
international mandates such as the Montreal Protocol, London Amendment, and the
George Bush Proclamation emerge to pose restrictions and elimination deadlines for the
usage of CFC based solvents. Many countries agreed to abide by the mandates and the
end of 1995 saw the elimination of production of CFC based solvents by many countries
including the United States. Less harmful solvents, such as HCFC based solvents have
also been blacklisted for elimination but have a slightly longer timetable for elimination.
7
8
1996 Contact East Supply Catalog, Page 178.
Lea, Colin. After CFCs. Electrochemical Publications, British Isles, 1992. Page 3.
10
2.4 CURRENT CLEANING OPTIONS AVAILABLE
Currently, since it is very difficult or impossible to obtain CFC-based solvents,
electronics assembly houses have been forced to adopt alternative cleaning methods.
Presently there exists five choices for manufacturers. Table 2 summarizes the options
available.
TABLE 2: Current Electronics Cleaning Alternatives
Aqueous
Aqueous w/sapponifier
Semi-aqueous
Pure water cleans
Relatively inexpensive,
Limited flux options,
assembly
least exposure hazards,
reduced cleaning ability,
good cleaning ability
waste effluent
Pure water with
Wide range of fluxes
Waste effluent
sapponifier cleaning aid
cleanable, good cleaning
added
ability
Pure water and alcohol
Wide range of fluxes
Some exposure (YOCs)
mixture
cleanable, good cleaning
hazards, waste effluents
ability
Solvents with reduced
HCFCs, or other
Excellent cleaning
Availability, cost, waste
ODP
chlorine based solvents
ability
effluent, ODP
No-clean
Benign soldering
No waste effluent or
Tighter process and
residues left after
associated costs,
material requirements
soldering
increased throughput
The choice of an alternative is subject to many factors such as the type of assembly,
cleanliness required, soldering materials used, budget, and many other factors.
All the options listed except for no-clean still have negative environmental effects
however. If cleaning is performed, waste effluent is generated either continuously or
through periodic purging of cleaner filtration systems. Others factors make cleaning
options less desirable also, such as cleaning system maintenance requirements, capital
investment, electricity usage and floor space expenses. A no-clean process in which the
electronics assemblies would not have to be cleaned at all would prove to be the best long
term solution.
11
CHAPTER 3: SOLDER PASTE MECHANICS
3.1 SOLDER PASTE BACKGROUND
In SMT, since soldering is used for both the mechanical and electrical connection,
it is imperative that an understanding of the mechanics of solder paste develop. Solder
paste is a thixotropic mixture which has a consistency similar to peanut butter. Its
thixotropic nature allows it to thin while under the pressure of shear from the squeegee
during the printing operation and be deposited through the stencil apertures. Once
shearing force is removed however, it immediately thickens and retains its shape. In this
thickened state the paste can then release from the stencil as it is raised away from the
stenciled circuit board.
A typical solder paste for printing consists of between 88 and 92 percent metal by
weight, or approximately 50% by volume. The solder metal is in the form of small,
round particles of the solder alloy suspended in a flux binding matrix. This combination
gives the paste the proper viscosity and rheological properties to allow it to be printed.
As such, a solder paste is a highly sophisticated formulation and offers a wide range of
possible formulations. 9 Figure 4 shows a pictorial representation of the major
constituents of solder paste.
Solder Paste
1
II Flux Binder
Solder
Alloy
I
I
s.:., I
IActivators
ISolvents and
Modifiers
FIGURE 4: Solder Paste Constituents
Hutchins, Charles L. Understanding and Using Surface Mount and Fine Pitch Technology. North
Carolina, 1995. Page 5.
9
12
3.2 SOLDER ALLOY
The solder material requirements for most electronic assemblies 'are met with an
alloy of tin and
lead~
Typically a ratio of 63 % tin and 37% lead is used although different
ratios or materials such as gold, silver, indium, or others are sometimes substituted to
provide variations in solder properties. The 63/37 alloy is a eutectic alloy, indicating that
it changes directly from solid to liquid at 183 degrees Celsius, and does not have a plastic
state. This common solder alloy was the type used for this study. Figure 5 shows the tin,
lead phase diagram which illustrates the different melting temperatures of various weight
combination ratios for the tin and lead.
350
300
250
200
TEMP
r'C>
150
100
50
Pb 10
20
30
40 50 60
TIN. W8GHT%
70
80
90
Sn
FIGURE 5: Tin Lead Phase Diagram 10
There are six separate phases depending upon the temperature and the ratio. There are
two crystalline forms of tin-lead commonly referred to as alpha and beta. These crystals
exist singularly at the temperatures and ratios shown in the phase diagram. 11
10
11
Hutchins, Page 73.
Hutchins, Page 74.
13
3.2.1 SOLDER ALLOY PARTICLE SIZE
Solder alloy particle sizes have variation within a paste. Particle size of a solder
paste is generally specified in terms of the size of a mesh that it will pass through. A
paste with a mesh of -200 to +325 (diameter range of 45 to 75 microns) means that it
would pass through a 200 mesh screen but not through a 325 mesh screen. 12 Generally, a
paste of -325 to +500 (Type 3) is used for stencil printing because it has empirically
shown to give the best compromise of results between standard pitch, fine pitch, and
ultra-fine pitch spacing. Type 3 solder pastes were the particle size used for this study.
3.3 FLUX BINDER
The flux binder is the most sophisticated ingredient present in a solder paste. It is
essential that the flux binder system be understood since issues with the resulting residues
are one of the largest reasons for cleaning or not cleaning a soldered assembly.
The flux binder system serves many essential roles in soldering without which
solder paste and surface mount technology would not be possible. It must suspend the
solder particles to give the paste its necessary rheological properties for printing. It must
also provide adequate ability for the paste to hold components once placed. Most
importantly, it must provide adequate fluxing action to remove the oxides on the solder
particles and base metals for soldering to occur. The major constituents of a typical flux
binder are the solids portion, solvents and rheological modifiers, and flux activator(s).
Detailed descriptions of the materials and issues involved with each of these components
for no-clean fluxes are discussed in Chapter 4.
3.3.1 FLUX BINDER SOLIDS PORTION·
As the name implies, the solids portion of the flux consists of non-volatile
material which remains on and around the solder joint after soldering procedures. The
solids portion provides many necessary functions in the flux binder matrix. The first
function which the flux solids performs is that it thickens the remainder of the flux
14
materials for maintaining rheological properties. The viscosity and make up of the binder
must be adequate to suspend the solder particles in order for the paste mixture to be
homogeneous. The solids portion melts at elevated temperatures below the solder alloy
melting point and coats the surfaces to be soldered, protecting them against re-oxidation
once it has been removed. When melted, the solder alloy "pushes" the flux solids to the
periphery of the solder joint. Once cooled, the flux solids re-solidify and remain around
the periphery of the solder joint.
3.3.2 FLUX BINDER SOLVENTS AND MODIFIERS PORTION
The solvents and modifiers in a solder paste serve the function of dissolving the
solids portion of the flux binder and combine to give it the proper viscosity and
rheological properties required. During the soak zone period of reflow, the solvents and
modifiers evaporate leaving behind the alloy, flux solids, and activator(s).
Depending on the formulation of the paste used, the solids portion of the flux
binder may contain some inherent ability to remove oxides from the solder particles and
base metals at elevated temperatures. Generally however, this fluxing action is not
sufficient to remove the oxides on most solderable surfaces. To solve this problem,
activator(s) are added to the flux binder system whose sole purpose is to remove
oxidation from the surfaces.
3.3.3 FLUX BINDER ACTIVATORS PORTION
The role of the activators that are added to fluxes has been previously stated.
Essentially, they are materials which are added to the flux to provide the majority of its
ability to remove the oxides from the solderable surfaces at elevated temperatures. The
materials that are added as activators vary considerably depending on the type of flux
being utilized. Generally, the activators present in a solder paste are ionic and consist of
acids and/or salts, which remove the oxidation through standard chemistry oxidation
reduction reactions. Being ionic, they dissolve readily in moisture, which affects the
12
Hutchins, Page 76.
15
surface resistivity of the assembly and ,leads to the potential to cause major corrosion and
electrical leakage problems if not cleaned from the assembly.13
13
Lea, Page 106.
16
CHAPTER 4: NO-CLEAN SOLDERING
4.1 NO-CLEAN BACKGROUND
The principle of not cleaning electronics assemblies after soldering is not new.
For many years huge quantities of inexpensive, low performance or short lifetime .
consumer electronics goods have been produced without cleaning. For higher reliability
electronic assemblies, the expense and procedural problems of cleaning have been
tolerated because at worst the effects may be catastrophic. 14
A no-clean flux is simply a flux which can be left on the assembly without
detrimental effects to the end use of the product. Rosin, a flux solid, has been used for
decades in fluxes and has been left on many electronic assemblies over the years. It has
even been shown that rosin provides many inherent properties which show little potential
for corrosion or electrical failure. This physical nature of rosin based flux residues has
been the cause of research towards new generations of no-clean flux formulations. Rosin
based fluxes tend to leave large quantities of residues behind, which can be frustrating to
automatic electrical test procedures and obscuring to visual inspections. This has lead to
significant research in to alternative chemistries to produce no-clean fluxes with little to
no solids portion remaining after soldering, and the development of fluxes which are resin
based instead of rosin based. Rosin is a distillate of naturally occurring pine sap whose
source has its own chemical fingerprint and variability. Resin on the other hand, is a
man-made substance which is more reproducible. 15 Generally speaking, no-clean fluxes
have the following characteristics:
-Low solids content
-Low residue
-Non-tacky surface hardness
-Lower fluxing ability
- Low potential for corrosion or electrical degradation 16
Lea, Page 295.
Lea, Page 296.
16 Lea, Page 297.
14
15
17
4.2 FLUX AND NO-CLEAN FLUX MATERIALS
The activity level, the ability to remove oxides from the soldering materials and
surfaces, varies depending on the type of flux. The most aggressive acid based fluxes are
too corrosive and consequently are not used in the assembly of electronics. For many
years during the height of CFC based cleaning, rosin mildly activated (RMA) fluxes were
the workhorses of most assembly shops. They offered long tack times, excellent
printability, and good fluxing action. Their relatively high quantity of residue levels were
not an issue due to their almost effortless clean up with CFC based solvents. Organic
acid water soluble fluxes (WS, OA) became and have remained popular since CFC
reduction began. They maintain a relatively high activation level and could can be
cleaned easily with plain water or water with a soponifier added. However, water soluble
fluxes exhibit significantly less work life than their RMA counterparts. All fluxes
available for soldering fall under one of the classes shown in Table 3.
TABLE 3: Classification of Soldering Fluxes By Main Ingredients 17
Resin
Rosin (Natural)
No Acivator, Halogen Activated, Non-Halogen
Activated
Resin (Man Made)
No Acivator, Halogen Activated, Non-Halogen
Activated
Organic
Water Soluble
No Acivator, Halogen Activated, Non-Halogen
Activated
Non-Water Soluble
No Acivator, Halogen Activated, Non-Halogen
Activated
Inorganic
17
Salts
With or Without Ammonium Chloride
Acids
Various Types
Alkalis
Amines, Others
Hymes, Page 54.
18
No-clean fluxes are taking more of a foot hold in the industry now that companies
are realizing their potential benefits. The solids percentage in no-clean fluxes are
reduced in order to decrease the amount of residual material which has the potential for
ionic reaction and thus potentially increase reliability. In addition, the activators are
much less aggressive than the previously mentioned fluxes in order to reduce the
possibility of ionic reaction. The chemical breakdown of these activators is proprietary
and thus impossible to specify here. As manufacturers push the solids and activity levels
of no-clean fluxes downward however, sometimes reflow in an inert atmosphere such as
nitrogen must be performed to compensate for the reduced fluxing levels. The nitrogen
helps to protect soldering surfaces from re-oxidation once removed which would
previously be accomplished by a higher volume of flux solids in the paste.
No-clean pastes are generally categorized, depending upon the type and amount of
flux solids, and the recommended reflow environment. Table 4 shows the generally
accepted classifications for no-clean solder pastes.
TABLE 4: Typical No-clean Specification Breakdown
Standard
::0:3.5%
Rosin
Air
Rosin/Resin
Air I Nitrogen
Resin
Nitrogen
OVERALL
WEIGHT
Low
3.5%::0: WEIGHT
::0: 1%
Ultra-low
1%::0: WEIGHT
4.3 NO-CLEAN ADVANTAGES
The elimination of a post solder cleaning step offers many potential advantages to
those processes which do require cleaning. A list of benefits include:
eElimination of environmentally damaging solvents, and other waste effluents
19
ePotentially, the complete elimination of all cleaning equipment and associated costs
such as capital investment, maintenance, floor space, energy usage, etc..
eCompatibility with non-washable components that would otherwise have to be
added post wash
eGenerally, a longer processing window due to the less hygroscopic nature of noclean fluxes
eIncreased throughput resulting from elimination of time used for a cleaning
operation
eElimination of concern about hard to clean areas such as under lower profile
components 18
All these benefits are very desirable for electronics assembly manufacturers. However, as
in any newer process there are concerns.
4.4 NO-CLEAN CONCERNS
Unfortunately, no-clean fluxes are not usually a direct drop in replacement for
cleaned fluxes. As previously stated there are many concerns which plague its
adaptability for a given process. Some of the concerns associated with no-clean fluxes
include:
eTighter manufacturing process control and incoming material solderability control
procedures must be adopted because no-clean fluxes generally have lower fluxing
activity than fluxes designed to be cleaned off a board
eTesting must be accomplished to ensure equal or better end reliability of the
manufactured product against previous manufacturing processes
eMaterial compatibility's and their effect on reliability must be ascertained
eAesthetic misconceptions resulting from the visibility of residues left behind must be
overcome
elf probe or bed of nails type testing is to be done, the residue's effect on the ability and
level of probes contamination must be determined
eIf required, effects on the ability of post assembly coatings (conformal coatings) to
adhere reliably to flux residues must be examined
eStrict process control on manufacturing handling must be implemented to ensure the
elimination of handling related contamination
eContaminants other than flux residues such as component residues, solder balls, and
PCB fabrication residues must be controlled and monitored
eCustomer acceptance and perception of quality must be addressed 19
18
19
Hymes, Page 5.
Hymes, Page 5.
20
While these are serious issues which must be addressed, if proper care and qualification is
completed, these factors can be overcome, even in a difficult manufacturing-setting.
21
CHAPTER 5: IMPLEMENTATION METHODOLOGY
5.1 INTRODUCTION
No-clean solder pastes would seem to be the ideal replacement for pastes that are
cleaned. However, they are not simply direct drop in replacements for current processes.
Proper evaluation, qualification, and education must occur before implementation should
be attempted. The main areas of concern when implementing no-clean have been
previously stated, although it should be noted that each point has a varying degree of
importance depending on the specific product and manufacturing setting. For example, it
would be easier for a manufacturer who has a dedicated automated assembly line to
implement no-clean than a contract manufacturer with a smaller throughput and larger
product mix. Issues such as handling contamination potential decrease with an increase
in the level of automation. Process parameters can be adjusted less frequently for a
dedicated production line; making process control simpler than on lines with constant
product changeovers. These differences though, do not change the impetus for all
manufacturing settings to change from environmentally deleterious cleaning methods to
friendlier techniques.
Some of the larger companies in Massachusetts have already converted to a noclean process. However many of the smaller contract manufacturing houses have had
difficulty converting immediately to a no-clean process and have opted to go with a more
forgiving non CFC based cleaning option such as water based cleaning. Many of these
companies lack the time, money and engineering resources to properly investigate,
qualify, and implement no-clean.
5.2 GENERAL IMPLEMENTATION STRATEGY
All implementation methodologies will vary, just as no two electronic assembly
companies are the same. Factors such as product type, product mix, volume, and budget
will greatly impact the implementation scheme. There are some general milestones and
rules that tend to stay the same however. Figure 6 shows a flow chart of what the
22
milestones for a general implementation plan may look like. This is obviously not the
only way to approach the problem, and should be taken in such context.
1
Make No-Clean
Decision, Set
Goals
y
2
Form
Implementation
Team
-.
...
4
3
Evaluate
Options, Define
Approach
Contact
Customers and
Suppliers
.J,
5
Decide on
Materials,
Process
y
y
IAssess ~PStrea m
Concerns
I
6
Qualify,
Evaluate, Iterate
~
I
14
I
I
Educate
Personnel
~
8
Implement Incomi ng
Material Audits
9
Specify
Process
i
y
Perform Real
World Testing
i
y
11
Get Final
Customer
Approval
.J,
12
Implement
Fully
I
y
13
Continue Audits,
I Process Control
!
!I
FIGURE 6: General Implementation Approach
23
1)
MAKE NO-CLEAN DECISION, SET GOALS
One of the most important steps in any no-clean implementation is the initial
decision to achieve a no-clean process. It must be decided at this initial step how serious
the effort will be and what the goals are. How imperative the change is will directly
influence the effort given. The most important point of this step to remember is- "the
success of implementation realized is directly related to the effort and time spent
evaluating and qualifying".20
2)
FORM IMPLEMENTATION TEAM
Forming a team to follow the process from start to finish is a very important point
to obtaining successful results. The team should consist of members of every aspect of
the business who have any input to the assembly process. This includes everyone from
upper management, quality control, material acquisition, process control engineers,
technicians, and assembly workers. If outside help is needed through consultants or other
professionals, it should be sought. Involvement should come from all members.
3)
EVALUATE OPTIONS, DEFINE APPROACH
As this will vary with each specific situation, it is possibly the most important
step of the entire implementation process. Research must be accomplished to understand
all options available. A realistic approach needs to be derived, along with a timetable and
cost benefit analysis. Which qualification measures that are relevant to the
manufacturing setting must be defined at this stage.
4)
CONTACT CUSTOMERS AND SUPPLIERS
Customers and suppliers must be involved at an early stage of implementation.
Customers must be contacted and not only be made aware of the intention to convert to
no-clean, but be asked to provide any specific requirements they may have. Suppliers
must also be informed as to what is acceptable and what is not. Because of the reduced
20
Interview with Dr. Brett E. Hardie, Hardie Consulting Co., Austin, Texas. 1/6/96
24
activity level of no-clean fluxes, incoming material solderability guidelines must be
generated.
5)
DECIDE ON MATERIALS, PROCESS
Once the previous milestones are completed, the decision of which materials and
process best suits the situation must be made. The results from the previous stages must
be assessed, evaluated, and a final approach be derived.
6)
QUALIFY, EVALUATE, ITERATE
The validation of the materials and the process must be tested to ensure that they
will be acceptable. Verifying the end product performance is one of the most important
steps during implementation.
7), 8) ASSESS AND CONTROL INCOMING MATERIALS
The extent, tests, and measures for controlling incoming materials for
solderability and residual fabrication residues must be addressed at this stage.
Verification that what is specified is indeed what is supplied is of the utmost importance
for successful implementation and maintaining process controls.
9), 10) SPECIFY PROCESS, TEST ON PRODUCT
Once evaluation is completed and a process is specified, it must be tested on an
actual product. Good laboratory results do not necessarily correlate to good real world
results.
11)
FINAL CUSTOMER APPROVAL
Once the materials and the process have been qualified and implemented
successfully, results must be supplied to customers for final approval and acceptance.
25
12), 13)
IMPLEMENT FULLY, MAINTAIN PROCESS CONTROL
These two steps go together intimately. Changes in process control,especially for
incoming materials, must be closely monitored through audits and changed as required.
14)
EDUCATE PERSONNEL
While this step is listed last in this list, it should never be considered last. Proper
education of personnel as to specific no-clean issues such as handling changes and visible
residues should occur in parallel throughout the implementation process.
5.3 RESEARCH IMPLEMENTATION METHODOLOGY
Tristar Technologies Incorporated was the electronic manufacturer chosen for
implementation. Tristar is located in Methuen Massachusetts and provides a full range of
services relating to printed circuit boards and electronics. They offer design, fabrication,
and assembly services to customers. Tristar employs a work force of approximately 250
employees on three shifts in a 120,000 square foot facility. They became ISO 9002
certified in 1995 and had gross sales of 26 million dollars in fiscal year 1996. Some of
their customers include Zenith Data Systems, Teradyne, Inframetrics, and Fostex
Research.
Currently they utilize a closed loop 100% aqueous cleaning system for assembly
in conjunction with an organic acid water soluble solder paste. While they have a desire
to convert a percentage of their throughput to no-clean, they have not had the time,
resources, or means to properly qualify and implement a no-clean process.
Many industry standards are available for the evaluation of no-clean solder pastes
for surface mount technology. While these standards help to define acceptability for
reliability issues, they do not offer much help in process implementation.
It was decided that multiple pastes from all major solder paste vendors would be
obtained and qualified against criteria specifically relevant to Tristars assembly setting.
The two most robust pastes which were found to provide the best relative results would
then be allowed to pass to process implementation.
26
For implementation, the parameter settings for the SMT assembly equipment
using no-clean pastes had to be found. To accomplish this, it was decided that design of
experiments (DOE) methodology offered the best alternative for isolating and optimizing
parameter settings for Tristars assembly line. Design of experiments (or matrix
experiments) was chosen due to its statistical approach and could be completed for ultrafine pitch, fine pitch, and standard pitch surface mount technology. After factors were
analyzed, the final result would be process parameter specifications which would yield
acceptable results for each case of standard, fine pitch, or ultra-fine pitch surface mount
technology. A flow diagram of the implementation methodology can be seen in Figure 7.
27
Obtain
Pastes
Qualify,
Evaluate
Pastes
f-1------..
Reject Pastes
y
Analyze
Tristar Stencil
Printer (DOE)
Analyze parameters for relative
contribution to output
Repeat experiments for variability
analysis
Quantify deposited brick width
IFull Process
I
DOE
L_ _
...
~_
,
Analyze full SMT process
Quantify reflowed solder joint
fillets
I
y
Specify
Process
i
!
,--~
Standard
Pitch
Fine Pitch
, Ultra-Fine
pitch
I
r---~
i Real World
I
Test
I
FIGURE 7: Research Implementation Methodology
28
CHAPTER 6: MATERIAL QUALIFICATION AND EVALUATION
RESULTS
6.1 INTRODUCTION
Ten standard residue no-clean soldering pastes from nine different manufacturers
were obtained. They were designated pastes A through J to retain anonymity. Standard
residue level pastes were used due to Tristars air reflow environment within the reflow
oven. While some of the properties, such as viscosity, varied slightly between pastes and
manufacturers, all pastes were off the shelf formulations.
Many standardized tests exist for qualifying pastes. The Institute for Packaging
and interConnection (IPC) has been responsible for generating the most widely adopted
industry standards at this time. Some of these evaluation measures were not attempted
for this study due to either lack of access to very expensive equipment or subjectively low
relevance when compared to Tristars specific requirements.
Evaluation tests that were adopted were designed based on research, suggestions
by experienced industry contacts, and Tristars specific requirements. It was decided that
performing decided upon qualification measures on all pastes and placing the results in to
a rating matrix for an overall relative performance rating offered the best method for
selection. Only the two highest ranking pastes would be allowed to proceed beyond the
initial qualification round. The Evaluation measures were broken down in to two
categories: pre-reflow and post- reflow qualifications. Where applicable a four hour
process window was assumed which is specific to Tristars manufacturing setting. This
was agreed upon by Tristar engineering support as a realistic production window for the
worst case lag between printing and reflow. Besides pastes, other materials which were
obtained to be used in the this study included test circuit boards for both standard, fine
pitch, and ultra-fine pitch SMT. A picture of each can be seen in Figures 8 and 9.
29
FIGURE 8: Tristar Donated Ultra-Fine Pitch 8MT Test Board
FIGURE 9: Topline Brand 8MT Test Board
30
Dummy electronic components were also obtained to mate to the land areas of each of the
test boards. For stenciling purposes, five and six mil thick framed electrofonned stencils
were obtained for the ultra-fine pitch case. Five and eight mil thick chemically etched
framed stencils were obtained for the fine pitch and standard pitch cases.
6.2 PRE-REFLOW PASTE EVALUATION RESULTS
6.2.1 APPEARANCE
Appearance is considered to be subjective, but is the quickest initial way to
distinguish differences between pastes. While appearance is not a necessary condition for
acceptability, it offers a quick initial way to find distinctions in pastes.
All ten samples were opened and stirred to ensure homogeneity. The pastes were
then inspected for consistency, air voids, smell, and any other noteworthy conditions.
While this is a subjective test at best, it did offer the initial groundwork to note
differences between pastes. The main differences between pastes that were observed
were in the viscosity and odor. It was not surprising to find differences here due to the
"off the shelf criteria" that was followed. Some pastes were noticeably thicker than
others. Some pastes emitted an unpleasant odor but after consulting operators who would
have to work with the pastes on a daily basis it was decided that the extent of even the
worst odors were acceptable. It was decided that all pastes were acceptable and would
receive a "Pass" for the rating matrix.
6.2.2 TACKINESS AND ADHESION
One of the major requirements of a solder paste is its ability to hold the
components placed on it through successive component placement operations without
allowing components to move or fall off until permanent mechanical solder joints can be
formed through the reflow process. The tackiness is one of the main requirements for
performance in order to accommodate the lag between paste deposition on the board and
reflow?! The amount of adhesiveness exhibited by the paste is a function of the
rheological properties of any given paste and the exposure time to the ambient
31
environment. A fresh paste possesses a greater adhesive ability than a paste subjected to
ambient conditions for a period oftime. As the solvents in the paste volatilizes, the paste
thickens and loses it adhesive ability.
The IPe standard test method for tack measurement is completed using a
dedicated tack tester. Essentially, the tack force is measured by bringing a stainless. steel
probe in contact with the paste and slowly removing it at a given rate while registering
the force required to remove it. The same procedure is repeated at various time intervals
to take in to account the tackiness relative to the rate at which the solvents in the paste
volatilize.
In the absence of a dedicated tack tester, an alternate method for measuring was
established. This procedure was based on one proposed by J. Hwang in her book "Solder
Paste in Electronics Packaging". The measuring technique entails printing a given pattern
on to a blank flat substrate and placing a component on the printed pattern. Once
intimate contact has been established between the two, the unit is inverted and the time
until the component drops off is recorded. This procedure is repeated at various time
intervals after the paste has b~en printed and subjected to ambient conditions to ascertain
the differences in tack relative to the drying rate of the paste. While this is a qualitative
measurement at best, it experimentally worked well to characterize performance. The
procedure was repeated hourly for a four hour period after printing. This was based on
the four hour process window which was assumed for Tristars' assembly process.
The results obtained were averaged and a relative scale for rating was used
between 5 and 1 where 5 was best and 1 was poor. A summary of the overall ratings for
each paste tested can be seen in Table 5.
21
Hwang, Page 50.
32
TABLE 5: SUbjective Rating for Paste Tackiness Ability
ABC
D
E
F
G
H
I
J
345
134
1
3
1
5
It is apparent that there exists a wide range of performance between pastes for tackiness
ability.
6.2.3 SLUMP
The slump of solder pastes is defined as the spread of deposited paste under its
own weight from factors such as temperature and humidity. This is obviously an
undesirable property which could lead to bridging of adjacent conductors on the circuit
board which would require rework. The mechanical and chemical formulation of a paste
determines its ability to resist slumping. No-clean pastes are less hygroscopic than water
soluble pastes and generally show much greater humidity resistance.
The IPC test method for slump evaluation (IPC-TM-650.2.4.35) entails printing
two different set patterns on to non-wettable substrates. The test specimen are then
subjected to specific environmental conditions and evaluated for the extent of slumping
which occurred. The IPC test method procedure may be viewed in Appendix B.
For this study, stencils with the specified IPC test apertures were not available so
the 6 mil thick electroformed UFP (15.7 mil pitch) stencil was substituted. By using this
combination (the lowest pitch and thickest print available), the worst case scenario for
Tristars assembly area was taken in to consideration. Five samples of test boards were
hand printed and subjected to the IPC specified conditions in the University of
Massachusetts at Lowell Electronics Manufacturing Laboratory environmental chamber.
The width of each printed pattern was monitored using a video microscope measurement
system. Virtually no slumping occurred and it was determined that all pastes provided
excellent slump resistance under these conditions, and would receive a "Pass" rating.
33
6.2.4 DISPENSABILITY
The dispensability of a paste through a syringe is an important property of solder
pastes. Dispensing set amounts of paste on to areas of the board which may have been
missed due to mistakes from stenciling or if a stencil is not available for use. There is no
IPe standard for rating dispensability so a method had to be developed. After consulting
with engineering support at Tristar it was determined that the following method would be
performed. The method entails loading syringes with a given paste, and applying a set
quantity of dots on to a substrate using a pneumatic dispensing station. The weights
before and after dispensing were recorded. The same procedure was repeated after the
syringes had been stored for a one week storage period in ambient conditions. The
weights between the two runs were compared, which would be a pointer for separation in
the paste along with dispensing ability. After a one week period there were no
w~ight
differences larger than 5% and it was concluded that all pastes performed adequately
enough to receive a "Pass" for the final rating matrix.
6.3 POST REFLOW EVALUATION MEASURES
6.3.1 WETTABILITY OF FLUX (FLUX EFFICACy)
The ability of each paste to properly wet and adhere to an oxidized copper surface
is important to determine the efficacy of the activators in the flux portion of a paste and
the overall solderability ofthe paste.
According to IPe test method 650.2.4.45, three 8
mil thick by 0.25 inch diameter solder bricks are to be printed on to an oxidized copper
coupon and reflowed. The IPe test method procedure may be viewed in Appendix B.
The results are then compared at lOX for any signs of dewetting, or other undesirable
effects. This test procedure was performed on equally oxidized copper coupons and it
was determined that all pastes except for paste "A" exhibited acceptable solderability and
wetting. Paste "A" showed severe signs of inability to overcome oxidation on the copper
coupon. An example of the dewetting found by paste A can be seen in Figure 10.
34
FIGURE 10: Paste "A" Dewetting Example
All pastes except for paste A received a "Pass" for the final rating matrix.
6.3.2 SOLDER BALLING
Solder balling is a phenomenon of solder pastes in which some of the tiny solder
particles do not coalesce with the remainder of the solder during reflow and are left
around the periphery ofthe solder joint. Normally, the solder balls are entrapped in the
residue where they remain harmless. However, they have the potential to break loose and
short out adjacent conductors which is obviously a very serious threat to reliability. This
becomes a large issue when no-clean is concerned because previously cleaning would
remove all of these solder balls, and residues.
Solder balling can potentially be caused by many factors. One of the main factors
causing solder balling is highly oxidized solder particles which fluxing activity can not
overcome, which essentially get "spit" out and do not reflow with the rest of the solder
joint. Another common cause for solder balling is an incorrect reflow profile which
evaporates the solvents in the flux binder too violently causing spattering of solder
particles. Yet another factor which may cause solder balling is material incompatibilities
between the printed circuit board surface, the solder alloy, and soldering residues.
35
According to IPC test method 650.2.4.43, a 0.25" diameter by 8 mil thick pattern
is reflowed on a non-wettable surface such as alumina and examined for solder ball
presence under lOX to 20X magnification. The same pattern is printed and allowed to be
subjected to ambient conditions and reflowed hourly for the four hour process window.
This is to examine the differences in solder balling relative to the paste solvents
evaporating for actual processing requirements. The IPC test method procedure may be
viewed in Appendix B.
The IPC test method was followed and the outcomes showed a wide variety of
results. A graph of the quantities of solder balls present versus time can be seen in Figure
11, and the numerical results may be viewed in Appendix A
ISolder Ball Evaluationl
35
.-
30
~ 25
.....PASTEA
-=
-W-PASTED
~
~
-'il-pASTE B
20
PASTEC
~
15
:g 10
-PASTEE
s.
~
=
-"'PASTEF
......PASTEG
00.
-PASTEH
5
-=-PASTEl
o
~ASTEJ
o
1
2
3
4
Time (Hours)
Figure 11: Graph of Solder Ball Test Results
The results showed that paste G exhibited completely unacceptable solder balling levels
along with paste F and H to a lesser extent. These vendors were contacted and ultimately
conceded that these lots exhibited poor solder ball resistance. This did not bode well for
confidence in their product consistency. The control metric which was used for the final
36
rating matrix was an average often solder balls. If the overall average for a paste
exceeded this, then it received a "Fail". An example of results obtained from paste G and
paste J can be seen in Figure 12. Note the significant differences between a good
performing paste versus a poorly performing paste.
Sample Solde.r Balling Results From Paste J (EXCELLENT)
Sample Solder Balling Results From Paste G (POOR)
FIGURE 12: Visual Comparison Between Solder Balling Results
37
6.3.3 ELECTRICAL CLEANLINESS (IONIC CONTAMINATION)
Ionic contamination represents one of the most significant factors causing
degradation and failure of electronic assemblies. Very small quantities of residual ionic
contamination can cause catastrophic failure of components and circuits under certain
conditions. Residual ionic contamination can cause surface electrical leakage, chemical,
galvanic, and electrolytic corrosion. 22 Two methods which exist for quantifying the ionic
contamination present which were used for this study include: resistivity of solvent
extract and surface insulation resistance measurements.
6.3.3.1 RESISTIVITY OF SOLVENT EXTRACT
The measurement of the concentration of ionic species left on an assembly is an
important pointer to the reliability of an assembly. As previously presented, ionic
residues are the chief source of electrical and corrosive properties of flux residues. IPC
test method 650.2.3.26.1 has specified a method of measuring such ionic residues. A
soldered assembly is placed in a bath containing a semi-aqueous solution (approximately
75% alcohol and 25% water mixture) which dissolves the ionic species on the assembly.
As the ions are dissolved, the resistance of the solution changes and can be recorded. By
knowing the area of the board, the solution volume, and the change in the solution
resistivity, the concentration of ionic residues can be equated to equivalent concentration
ofNaCl. The IPC test method procedure may be viewed in Appendix B.
There are many types of equipment available to accomplish this measurement.
For this study, an Alpha brand Omega Meter was used. The unit utilized for this study
can be seen in Figure 13.
22
Omega Meter Users Manual, Page 1.
38
FIGURE 13: Alpha Brand Omega Meter
Four boards of each paste utilizing the Topline boards were stenciled and reflowed using
each manufacturers recommended reflow profile. The Topline test boards were used for
their increased solder quantity density present on the surface of the substrate. The tests
were run four times for result confidence and the resulting contamination quantities found
were averaged. A summary of the results obtained can be seen in Table 6.
39
TABLE 6: Experimental Solvent Extract Resistivity Results.
A
23.8
F
B
4.9
P
e
4.9
p
D
16.6
F
E
12
F
F
14.1
F
G
12.6
F
H
1.3
P
I
6.5
P
J
9
P
According to the IPe standard test method, an equivalent contamination exceeding 10
micrograms per square inch is 'considered a failure. It is interesting to note that half of the
pastes tested failed this test.
6.3.3.2 SURFACE INSULATION RESISTANCE
Surface insulation resistance (SIR) was the most lengthy and difficult evaluation
measure that was accomplished for this study. The measurement of cleanliness
previously described depends on a change in the property of a solvent used to remove
contamination from the assembly. The solvent extract method was carried out on the
assumption that the contamination is detrimental to the electrical functioning of the
circuit and that the disolvability rate of each residue was very similar in the semi-aqueous
solution. Surface insulation measurements attempt to directly link the contamination to
the reliability of a circuit. 23
SIR tests measure the surface resistance between two adjacent conductors on a
circuit board under artificially hostile conditions over at set time period, the principle
40
being that ionic contamination on the board surface will lower the surface resistance and
degrade electrical performance. There are potential drawbacks to this method and it
should be taken as a complementary measurement to solvent extraction, although it does
offer the a better link to product reliability.
6.3.3.2.1 SIR THEORY
Surface resistivity has the unit of ohms per square. If a square is considered
having length X on the conductive side and length Y on the insulating side (See Figure
14), then the electrical resistance measured between the two increases linearly as Y
increases but decreases linearly as X increases. Therefore the measured resistance is not
affected by the size of the square, only the number of squares present.
FIGURE 14: Adjacent Conductors For SIR Testing
IPC test method 650.2.6.3.3 states that a test coupon with a known trace pattern
be subjected to 85% relative humidity, 85 degrees Celsius, while under a bias voltage of
50 Volts DC. The electrical resistance is measured using either a pico-ammeter and
Ohms law, or directly with a high magnitude resistance meter at various time intervals
over the course of a week under a measurement voltage of -1 00 Volts DC. The
23
Lea, Page 19.
41
resistance per square can found and should not drop below 109 Ohms/Square. The IPC
test method procedure may be viewed in Appendix B.
6.3.3.2.2 EXPERIMENTAL SETUP
In the absence of an environmental chamber which could obtain the extremely
harsh conditions specified, an alternate method was used. A constant humidity solution,
potassium bromide (KBr) dissolved in de-ionized water, was placed in a sealed oven at
85 degrees Celsius. KBr dissolved in de-ionized water at 85 degrees has the property of
maintaining a relative humidity of 80% in a closed area once equilibrium is achieved.
This five percent difference between the specified and obtainable humidities was deemed
acceptable.
The IPC test method specifies that an interdigitated comb pattern be used to
measure the SIR. In the absence of obtaining such coupons, a slightly older style military
"Y" coupon was used. An example of the test coupon used can be seen in Figure 15.
Note the exposed parallel conductive traces.
Exposed Parallel Traces
FIGURE 15: Military "Y" Coupon Used for SIR Testing
The test coupons were reflowed with each no-clean paste and soldered to coaxial
cable to eliminate electrical noise. A coupon was also reflowed using Tristars current
water soluble paste and cleaning equipment for a baseline indicator. The test coupons
were then mounted to a fixture, placed in the chamber, and connected to a variable power
supply and a pico-ammeter. A pico-ammeter was used in the absence of availability of an
adequate ohmmeter. The experimental setup used can be seen in Figure 16.
42
Note Constant Humidity Solution on Left Side of Oven and Test Coupons on Right Side
FIGURE 16: Experimental SIR Set Up
6.3.3.2.3 EXPERIMENTAL RESULTS
The experimental set up that was utilized worked well to record the insulation
resistance degradation over the course of the experiment. Measurements were taken once
a day for the duration of the seven day test. All reflowed coupons maintained an
insulation resistance above 109 ohms/square except for paste "A", which stayed below the
pass/fail criteria for the final four of the seven days. This SIR failure correlated with the
poor performance of paste "A" for solvent extract resistivity measurements. All pastes
except for" A" received a "PASS" for the rating matrix.
6.4 SIGNIFICANT TESTS NOT PERFORMED
While many evaluation measures were accomplished, there exists literally dozens
of potential qualification measures which can be performed on a solder paste. For the
43
purposes of this study, the following tests would have been run but could not for varying
reasons.
6.4.1 ELECTRO-CHEMICAL-MIGRATION
The electro-chemical-migration of metals in electronics has been recognized for
over 30 years. The phenomenon causes electrical shorts as two adjacent conductors are
bridged by a growth of metal dendrites. A metal which can form ions in the presence of
moisture, and have mobility under electrical voltage potential can exhibit electromigration. Therefore the requirements for electro-chemical-migration are: voltage
potential, temperature, time, and electrolyte (humidity).24
Under favorable conditions, the metal ions will literally migrate from cathode to
anode and form dendrite like structures between the two electrodes. While this is an
important test to be completed, the IPC specification calls for a test with a duration of
thirty days which was not feasible for the purposes of this study.
6.4.2 PRE-REFLOW CORROSION
Many tests are available to qualitatively and quantitatively rate the COITosivity of a
paste prior to reflow. In the absence of the equipment and materials to properly perform
this test, it was assumed that each manufacturers specification of acceptability for thIS
measure would be satisfactory.
6.4.3 HALIDE PRESENCE
The significance of determining the quantity of halide present in a solder paste
relates to its historical use as an activator. For many years, halides (fluorides, chlorides,
bromides) were used as activators in pastes that were cleaned because they exhibited
excellent fluxing action. They presented no threat to product reliability as long as they
were cleaned off the board relatively soon after soldering. Unfortunately, they react with
lead, water and air to produce a very corrosive chemical reaction which makes them
detrimental for no-clean usage.
44
While manufacturers of the pastes examined claim a "halide free" product,
interviews with industry experts suggested that small amounts of halide are still
sometimes used. Due to the complexity and difficulty of quantifying halides, this test
was not performed.
6.5 PASTE EVALUATION CONCLUSIONS
The original goal of the qualification and evaluation round was to find the best
performing pastes relative to requirements for Tristars specific assembly environment. It
became apparent during testing that there existed distinct differences between the
different pastes performance despite manufacturers claims of meeting or exceeding
similar specifications. The results for each qualification were input in to the rating
matrix and can be seen in Table 7.
TABLE 7: Paste Evaluation Rating Matrix
APPEARANCE
2
2
1
000
1
2
o
2
DISPENSING
P.
P
P
P
P
P
P
P
P
TACKINESS
3
3
5
134
1
3
1
5
SLUMP
P
P
P
P
P
P
P
P
P
P
10
EFJ;<ICACY
F
P
P
P
P
P
P
P
P
P
10
SOLDER BALL
P
P
P
P
P
F
F
F
P
P
10
SOLVENT
F
P
P
F
F
F
F
P
P
P
10
F
P
P
P
P
P
P
P
P
P
10
P
5
RESISTIVITY
SIR
* Selected Best Performing Pastes
24
Hwang, Page 319.
45
Relative weights for each test were generated by discussion with Tristar engineering to
rate the relative importance of each quality for its manufacturing setting. Weights for the
relative importance of each qualification measure were multiplied by the response found
and totaled. A "Pass" was considered a one, while a "Fail" was treated as a zero.
Pastes C and J were found to have the best overall performance for the tests
performed and would be the only pastes to pass beyond the qualification round for further
analysis. The tabular rating matrix method worked well to scale the pastes for robustness
required for Tristars assembly line.
46
CHAPTER 7: STENCIL PRINTING ANALYSIS
7.1 INTRODUCTION
Being the initial procedure in the surface mount technology assembly process,
stencil printing is generally considered to be the most significant process step affecting
defect variability in the end product. The considerable number of adjustable factors
involved with stencil printing require that a systematic approach to analyzing and
understanding consequential factors affecting the response of the system be
accomplished. Design of experiments (DOE) or matrix experiments is one such
technique for analyzing and controlling an industrial process.
A matrix experiment consists of a set of experiments in which the settings of
various process parameters are changed and the corresponding results obtained are
analyzed statistically. This statistical analysis has many beneficial inherent properties
such as parameter contribution analysis, interaction examination, and process
optimization techniques. For these reasons, a DOE approach to analysis of the stencil
printing process at Tristar was chosen as the bes~ way to identify the significant
parameters in obtaining an acceptable solder paste print.
7.2 STENCIL PRINTER EQUIPMENT
Tristar utilizes a DEK brand model 249 semi-automatic stencil printer for its
printing process. It is an exceptional machine with features such as vision alignment,
fully adjustable tooling table, and repeatability rated at +/-0.0004 inches. The DEK 249
is used on a daily basis and could be used in almost any manufacturing setting with
variants available for conveyorized lines. The DEK 249 can be seen in Figure 17.
47
FIGURE 17: DEK 249 Semi-Automatic Stencil Printer
While some prototype jobs are still hand printed, the overall majority are run on the DEK
printer for better result consistency and reduction of defects caused by human error.
7.3 STENCIL PRINTER PARAMETER SELECTION FOR ANALYSIS
Before the DOE matrix could be designed, the scope of the experiment had to be
addressed as to how many parameters and levels would be examined. To help determine
the important factors for stencil printing, a brainstorming session was held with Tristar
engineers and operators who work with the machine on a daily basis. A list of potential
output affecting contributors was generated and categorized. The categorized list was
then input into a cause and effect diagram for improved visual interpretation. The cause
and effect diagram produced can be seen in Figure 18.
48
Stencil
Printer
i
Environment
Separation Speed
'<4-_ _--,-Tem peratu re
Squeegee Speed __
'<4----Snap Off
Sq ueegee M a terial
__
___Squeegee Pressure
Alignment_.....,.
'<4----H u m id ity
i
Optirn urn
Stencil
Print
Sold e r Paste ---"----..,..-~---w
Aperture
Dimensions
....._--.;;:::-------"''------_ _-..::::-_,S te n c it
Land
Roughness
PCB
_\~_ _---."
Thickness
Type
COPla~arity
Material
Supplies
FIGURE 18: Cause and Effect Diagram of Potential Factors Affecting Stencil Printing
Since a great deal of knowledge and experience with the stencil printer existed at Tristar,
it was decided that a three or higher level experiment was not required and a carefully
chosen two level experiment would provide adequate process modeling. After further
discussions with Tristar personnel, it was felt that the number of factors available in an
L8 orthogonal array presented an adequate number to be studied. What was believed to
49
be the five most significant factors along with two interactions between factors were
chosen for examination. Levels for the factors were carefully chosen so that the levels
represented a realistic operating range for the Dek 249 stencil printer. The chosen factors
and their levels can be seen in Table 8:
TABLE 8: Selected Parameters and Levels
SQUEEGEE PRESSURE(RELATIVE SETTING)
SQUEEGEE SPEED
INTERACTION: PRESSURE AND SPEED
SQUEEGEE MATERIAL
INTERACTION: PRESSURE AND MATERIAL
STENCIL THICKNESS
B
PASTE
J
Of the many potential output metrics, the critical parameter of deposited brick width was
chosen as the output characteristic to be quantified. Unacceptable deposit widths can
potentially cause defects due to reasons such as inadequate or excessive solder volumes
when components are soldered. For measurement of the deposited solder paste width, a
video measurement system was utilized. A 25 mil wide 40 mil pitch land pattern was
chosen for analysis on the Topline test boards, which accounts for a good representation
of the majority of the component pitch assembled at Tristar. The test area on the Topline
test boards can be seen in Figure 19.
50
Note: Surrounding Board Area Truncated in Figure
FIGURE 19: Test Pattern Area on Topline Test Boards for Stencil Printer Analysis
The chosen factors for analysis were allocated to a modified linear graph for
assignment to the proper columns in the Taguchi L8 orthogonal array_ The modified
linear graph can be seen in Figure 20.
INTERACTION
FACTOR #3
~
SQUEEGEE SPEED
FACTOR #2
•
•
SQUEEGEE
STENCIL THICKNESS
PASTE TYPE
PRESSURE
FACTOR #6
FACTOR #7
FACTOR #1
INTERACTION /
SQUEEGEE MATERIAL
FACTOR #5
FACTOR #4
FIGURE 20: Linear Graph of Factors for Column Assignment in L8 Array
51
The L8 array with chosen parameters, levels, and column assignment can be seen in
Table 9.
TABLE 9: L8 Array with Chosen Parameters and Levels
FACTOR A
2
2
20
8
J
3
2
60
8
J
4
2
60
5
B
5
12
20
5
J
6
12
20
8
B
7
12
60
8
B
8
12
60
5
J
7.4 DESIGNED EXPERIMENT ANALYSIS AND RESULTS:
The proper level settings for each experiment were set and run with three
repetitions for each experiment. The target value was 25 mils, which is the width of the
aperture in the stencil. Stencil wiping between runs was done to ensure consistency
between trials. The experimental results obtained can be seen in Table 10.
52
TABLE 10: Experimental Results
1
23.86 mils
23.33
24.3
2
24.3
23.91
24.08
3
22.96
22.73
23.64
4
22.37
22.81
22.94
5
23.38
23.33
23.42
6
25
24.83
24.96
7
22.38
22.28
22.72
8
25.58
24.83
25.05
7.4.1 AVERAGE ANALYSIS
The results obtained were input in to a DOE analysis software package (ANOVATM) and analyzed for contribution to the average effect. The results obtained for the
primary factor effects on the average can be seen numerically in Table 11 and visually in
Figure 21 below.
TABLE 11: Average Effects of Primary Factors
F
STENCIL THICKNESS
23.77
23.65
G
PASTE
23.48
23.93
53
Primary Factor Average Effect
Stencil Printer Analysis
24.4
24.2
....t.J
~ 23.8
w
Q)
0)
E
Q)
>
<I:
23.6
23.4
23.2
A1
A2
81
D1
82
D2
F1
F2
G1
G2
Factor Levels
FIGURE 21: Plot of Primary Factor Average Effects
The plots of the average effects of the primary factors show that the impact on the output
characteristic due to the different factors varies. The effect of parameter D (squeegee
material) shows the largest influence on the average brick width. Other parameters such
as F (stencil thickness) show much a smaller influence.
The effects of the interactions that were studied can be seen in Appendix A. The
response between all the studied interactions are not parallel indicating that interactions
are indeed present. The significance of the interaction analysis shows that the squeegee
pressure, speed, and material parameters are not isolated factors when analyzed for their
affects on the average. When one parameter is changed it will affect the others. This
must be taken in to account when adjusting these parameters when optimization is
attempted.
To numerically quantify the differences in the percent contribution to the output
metric, an analysis of variance (ANOVA) on the average results was completed. The
ANOVA table generated can be seen in Table 12.
54
TABLE 12: ANOVA Results for Average Data
1
1
1
1
1
1
1
1.78
2.94
0.75
6.33
7.2
0.08
1.23
1.78
2.94
0.75
6.33 46.86
7.2 53.3
0.08
1.23 9.1
Error
18
2.43
0.14
Total
23
21.93 0.95
Parameters which exhibited less than five percent contribution to the output effect were
pooled as error (noise). A summary from the ANOVA table of the percent contributions
of the factors found significant can be seen in Table 13.
TABLE 13: Percent Contribution of Factors Affecting Average
SQUEEGEE PRESSURE
7.5%
SQUEEGEE SPEED
12.8%
SQUEEGEE MATERIAL
28.3%
INTERACTION: PRESSURE & MATERIAL
32.3%
PASTE TYPE
5%
NOISE (Error)
14.1%
The fact that all but 14.1% of the response of the system could be accounted for
was excellent, and showed that the process model chosen to analyze the system worked
well.
7.4.2 VARIABILITY ANALYSIS
To analyze the effect of the factors on the variability of the output characteristic, a
Taguchi signal to noise analysis was performed. The results obtained from performing
55
the experiment were input in to the Taguchi signal to noise equation:
SIN = 10 log lin ((Vm - Ve) / Ve)
Where: Vm = (Iy)2 / n
Ve = (Iy 2 - Variance of Error) / (n - 1)
y = Result Obtained
n = Number of Experiments Completed
The signal to noise values calculated for each experiment can be seen in Table 14.
TABLE 14: Calculated SIN Values for Results
1
33.82
2
41.82
3
33.78
4
37.62
5
54.29
6
48.96
7
39.77
8
36.29
A level analysis for the SIN data was performed using the ANOVA-TM software
package. The results obtained for the primary factor effects on the variability of the brick
width can be seen below.
56
TABLE 15: Variability Effect of Primary Factors
F
STENCIL THICKNESS
40.5
41.1
G
PASTE
40.04
41.54
Primary Factor Variability Effect
Stencil Printer Analysis
46
45
44
43
~
42
~ 41
z 40
en
39
38
37
36 ,
Ai
A2
81
82
01
D2
F1
F2
G1
G2
Factor Levels
FIGURE 22: Plot of Primary Factor Variability Effect
The plots of the effect on the variability of the output characteristic show that similar to
the average analysis, some of the factors exhibit larger influence on the variability than
others. Squeegee pressure and speed for example exhibited larger contributions than the
other factors.
57
The effects of the interactions studied can be seen in Appendix A. Similar to the
average analysis, both interactions studied exhibited the presence of an interaction
between factors. The squeegee pressure and material interaction showed more impact on
the variability than the pressure and speed interaction.
An ANaVA analysis was completed to find the percent contribution of the factors
to the variability of the brick width. The ANOVA table generated for variability analysis
can be seen in Table 16.
TABLE 16: ANOVA Results for SIN Variability Data
D
E
F
G
1
1
1
1
1
1
1
Error
3
6.33
2.11
Total
7
379.34
54.19
A
B
C
130.33 130.33
123.47 123.47
65.87 65.87
1.15
1.15
53.34 53.34 25.28
0.66
0.66
4.52
4.52
Again, parameters which exhibited less than five percent contribution to the output effect
were pooled as noise factors. A summary of the percent contributions of the factors
effecting variability can be seen in Table 17.
TABLE 17: Percent Contribution of Factors Effecting Variability
SQUEEGEE PRESSURE
33.8%
SQUEEGEE SPEED
32%
INTERACTION: PRESSURE & SPEED
16.8%
INTERACTION: PRESSURE & MATERIAL
13.5%
NOISE (Error)
3.9%
58
The process model chosen to analyze the system worked well, accounting for all but
approximately 4% of the response of the system.
7.5 STENCIL PRINTER ANALYSIS CONCLUSIONS:
The designed experiment approach to analyzing the surface mount technology
stencil printing process worked well to analyze the important factors affecting the
response of the system. Analysis of variance calculations led to the interesting result that
parameters which effect variability do not necessarily effect the average. This is clearly
demonstrated in Figure 23, which compares the percent contribution of each factor to the
output characteristic for variability and average results.
Factor Contribution To Output Charac1eristic
III Variability
ClAverage
o
5
10
15
20
25
30
35
Percent Contribution
FIGURE 23: Comparison of Parameter Contribution Effecting Average and Variability
Careful selection of the parameters and factor levels led to a model in which 86
percent and 96 percent of the response for average and variability respectively could be
accounted for.
59
CHAPTER 8: FULL SMT PROCESS ANALYSIS
8.1 INTRODUCTION AND APPROACH
The next step in the research implementation methodology was to perform a
designed experiment on the entire SMT process line at Tristar. After analysis, it would
then be possible to specify the optimum parameter settings for the process based on the
results found. To accomplish this, the scope, parameters, and the layout of the
experiment had to be determined. Based upon the analysis of the stencil printing results
found for the Dek 249 stencil printer previously and discussions with Tristar engineering,
the layout of the experiment was determined.
In order to analyze the three different lead spacing pitches to be examined, it was
decided that three separate designed experiments would be performed. Here, the idea was
to take the results found for each examination and to combine them mathematically to
specify process setting guidelines for all possible pitch mixes assembled at Tristar.
8.2 EXPERIMENTAL PARAMETERS AND LAYOUT OF MATRIX ARRAY
It was decided the standard Taguchi L8 orthogonal array that was used for the
stencil analysis would be employed again. This array and not a larger one with more
levels was chosen because of the substantial increase of processing steps, increase in
process parameters, and the fact that three different experimental arrays had to be run.
With the expansion of the array parameters extending to investigating the full process,
careful consideration had to be given to which parameters would be examined or held
constant and what the measured output would be.
Since the solder joint is responsible for both the mechanical bond and the
electrical connection, it was the criteria which was decided on for the characteristic
measured output metric. Industry standards, such as the IPe and Military specifications
have been established as guidelines of acceptability for such criterion. Based upon these
guidelines, a rating system was established to numerically assess the quality of the solder
joints that were formed resultant from each experiment trial. Fillet geometry was decided
60
upon as the qualification metric. The fillet shape geometry rating followed the guidelines
shown below in Figure 24, where a four (4) was the target condition.
L
RATING VALUE: 1
RATING VALUE: 3
RATING VALUE: ~
RATING VALUE: 7
FIGURE 24: Rating Criterion for Solder Joint Fillet Geometry
To aid inspection a 20X microscope was used. Cross sectioning was also accomplished
to properly relate outside inspection ofthe solder joint to the actual cross sectional
appearance of the fillet.
8.3.1 STENCIL PRINTER PARAMETERS
The knowledge gained from the stencil printing analysis helped to determine
which stencil printer parameters would be analyzed in the full process array. However, it
61
must be noted that the output metrics quantified between the stencil printer analysis
(deposited brick width) and the full process analysis (reflowedjoint appearance) do not
necessarily have a direct relationship. Because of this, some parameter modification was
necessary. Based upon the knowledge gained from the stencil printing analysis, the
following stencil printing related parameters were included in the full process
experimental array:
-Squeegee Pressure
-Squeegee Speed
-Interaction of Squeegee Pressure and Speed
-Stencil Thickness
-Interaction of Stencil Thickness and Squeegee Pressure
-Paste Type
Stencil thickness was added to the factors being examined despite showing up as not
significant from the previous printer analysis. This was done because the thickness of the
stencil greatly affects the volume of paste deposited and consequently the shape of the
solder joint fillet. Use of a metal squeegee was selected as the squeegee type to be used
exclusively because of the high effect on the results found from the stencil DOE analysis
previously. Stencil wiping and cleaning was held constant throughout the entire
experiment run.
8.3.2 PICK AND PLACE PARAMETERS
, Tristar utilizes a Mydata bran~ TP9-UFP pick and place machine for component
placement. It is considered throughout the industry as one of the best machines for a nondedicated assembly environment. Its flexible design permits virtually all relevant
parameters to be adjusted. For this investigation, it was decided that the placement
pressure of the components in to the paste would be the only factor introduced to the
array for the pick and place process step. All other adjustable factors would be held
constant. The Mydata TP9-UFP can be seen in Figure 25.
62
FIGURE 25: Mydata TP9-UFP Pick and Place Machine
8.3.3 REFLOW OVEN PARAMETERS AND OTHER FACTORS
Tristar uses a Heller 1500 forced convection reflow oven for solder paste reflow
in their SMT manufacturing line. The oven has five heating zones, each with both top
and bottom separately adjustable heater controls. The Heller 1500 convection reflow
oven can be seen in Figure 26.
FIGURE 26: Heller 1500 Forced Convection Reflow Oven
63
The manufacturers recommended thermal profiles for each of the two pastes that were
advanced to this point from the qualification round were matched and stored in the ovens
memory. The profile for whichever paste that was specified to be used in any experiment
trial was recalled and used only for that given paste. For the purposes of this experiment,
the reflow oven factors were therefore considered held constant. Other factors that were
considered held constant were the environmental conditions in Tristars manufacturing
area. Ambient conditions were set to 65 degrees Fahrenheit and kept within a range of
between 25 and 75% relative humidity. Since there are no humidity controls in Tristars
assembly area, the relative humidity was monitored so that it stayed between these levels.
Once the factors for the process were determined, assignment in to the array had
to be determined. Figure 29 below shows the linear graph of the factors for the array.
INTERACTION
STENCIL THICKNESS
FACTOR #3
FACTOR #2
•
•
SQUEEGEE
P.P. PLACEMENT
PASTE TYPE
PRESSURE
PRESSURE
FACTOR #7
FACTOR #1
FACTOR #6
INTERACTION/
SQUEEGEE SPEED
FACTOR #5
FACTOR #4
FIGURE 27: Assignment of Parameters to Array for Full Process Analysis
The levels for each factor that were chosen can be seen in Table 18 below.
64
TABLE 18: Selected Parameters and Levels for Full Process DOE Analysis
B
Stencil Thickness
C
Interaction of 1&2
D
Squeegee Speed
E
Interaction 1&4
F
P.P. Placement Pressure
G
Paste Type
B
J
* A 6 mil thick stencil was substituted for the ultra-fine pitch experiment array
** mN: milli-Newton unit of force
8.4 STANDARD PITCH RESULTS AND DISCUSSION
The standard pitch DOE was run using the layout discussed previously for four
repetitions. The components utilized were 1210 sized passive chip resistors mounted on
their corresponding land patterns on the Topline test boards. The test site can be seen in
Figure 30 below.
Note: Surrounding Board Area Truncated in Figure
Test Pattern Area Used for Standard Pitch Measurement
FIGURE 28: Test Site Used for Standard Pitch Measurement on Topline Test Boards
65
The results between the four repetitions were the same which eliminated the possibility of
analyzing variability. The results found can be seen below in Table 19.
TABLE 19: Results for Standard Pitch DOE
~.~I,
. ,;;;>...... 0:;,
....
;'j;,
> ,.,.'
1
3
2
3
3
5
4
5
5
3
6
3
7
5
8
5
Plugging in the results to an analysis of variance table yields the result that the difference
in fillet shapes were entirely due to the stencil thickness. This is most likely due to the
process being robust enough at 50 mil pitch that the other process parameter levels were
not set wide enough to affect the output. An example of a cross-section and its
corresponding rating can be seen below in Figure 29.
FIGURE 29: Cross-Section of Standard Pitch Lead; Rating = 3
66
The conclusion for the case of 50 mil and above pitch is therefore that the' process is
robust enough to yield acceptable results for any setting within the range of the settings
tested.
8.5 FINE PITCH ANALYSIS AND DISCUSSION
The fine pitch DOE was run using the layout and with four repetitions also.
The components utilized were 30 mil pitch 84 leaded quad-flat-packs mounted on their
corresponding land patterns on the Topline test boards. The test site can be seen in
Figure 31 below.
Note: Surrounding Board Area Truncated in Figure
Test Pa ern Area Used for Fine Pitch Measurement
FIGURE 30: Test Site Used for Fine Pitch Measurement on Topline Test Boards
Similarly to the results found for the standard pitch experiment, there was no variation in
the results between repetitions. So once again, no variance analysis could take place.
The results found can be seen in Table 20.
67
TABLE 20: Experimental Results for Fine-Pitch DOE
'he.:
">
4
•..••
,
"
,>
en.
• . ;.'1"\
;
....
1
4
2
3
3
5
4
5
5
3
6
3
7
6
8
5
An example a cross-section and its corresponding rating can be seen below in Figure 31.
FIGURE 31: Cross-Section of Fine Pitch Component Lead; Rating
=6
An ANOVA analysis was completed for the results found using the ANOVA-TM
software and can be seen in Table 21.
68
TABLE 21: ANOVA Results for Fine Pitch Process DOE
D
y
1
0.5
0.5
E
y
1
F
y
o
o
o
o
G
y
1
0.5
0.5
The results show that the stencil thickness has an unusually large contribution on the
output effect. The fact that this factor showed up as much more significant than the other
factors could be explained by the levels of the other factors not being set wide enough or
the stencil thickness levels being too wide as to swamp the other effects. Besides stencil
thickness, squeegee speed and paste type were found to be marginally important but were
discounted due to having slightly less than a 5% contribution. The effect of the paste
type could be explained by the difference in the two pastes viscosity. The effect of the
squeegee speed is possibly explained by the fact that as the squeegee moves across the
apertures in the stencil, paste deposition is affected by the rheology of the paste and the
squeegee speed to accommodate deposition. With the large bias of the results towards
the stencil thickness, it is impossible to analyze the average effects with any confidence.
Therefore, the average of the levels except for the paste type and stencil thickness were
used for the recommendations for optimum level settings. For the stencil thickness, the 8
mil thick was chosen because it would yield a more robust solder joint. The paste chosen
was Paste J because it had a thinner viscosity than paste B and seemed to print slightly
better than paste B.
69
8.6 ULTRA-FINE PITCH RESULTS AND DISCUSSION
The DOE was set up and run for the ultra-fine pitch pattern on the Tristar donated
test boards. Using the levels set up in the design, no acceptable solder joints resulted.
Virtually all leads after reflow were bridged across with solder. It is most likely that this
was caused by unacceptable printing resolution during the stencil printing stage of the
processmg. The poor printing could be resultant from many factors such as:
-Incorrect level settings
- Incorrect stencil aperture design
-Stencil printer inherent limitations
-Other factors
Because of the experimental array and the way it was designed, the factors could not be
changed after this was found.
8.7 FULL PROCESS DOE CONCLUSIONS
Despite the fact that unacceptable results were found for the ultra-fine pitch
condition, good results were obtained for the fine-pitch and standard pitch conditions.
Because the standard pitch results were found to be so robust, it can be concluded that the
optimum parameter settings are the average of the fine-pitch experiment setting levels.
For the purposes of this experiment, it is concluded that acceptable ultra-fine pitch results
were unobtainable for the levels chosen and that further research in to this area would
have to be completed if ultra-fine pitch patterns were to be accomplished properly.
8.8 CONFIRMING EXPERIMENT RESULTS AND DISCUSSION
To confirm that the results found for the full process DOE would work for an
actual product, a confirming experiment was completed. One of Tristars customers,
Clearpoint Research in Milford Massachusetts, was contacted and agreed to allow one of
their product types to be soldered using the no-clean process. The product was a
specialized memory module which utilized capacitor mounting under integrated circuits.
Because these areas are very difficult to clean, this particular product lent itself very well
70
to no-clean technology. A picture of the un-populated circuit board can be seen in Figure
32 below.
Note Capacitor Mounting Areas Under All Integrated Circuit Mounting Lands
FIGURE 32: Real World PCB Used for Confirming Experiment
The equipment was set to the average levels recommended from the fine-pitch DOE. The
assumption here was that the requirements for the fine pitch components on the board
would be targeted and the standard pitch components on the board would be robust
enough to yield acceptable results.
Thirty PCB's in total were run through the process and analyzed afterwards. The
results showed excellent solder joints formed that were all acceptable. Some integrated
circuits were removed to facilitate inspection of the capacitors below them. Careful
examination under 20X magnification revealed that they too soldered very well with good
fillets. Clearpoint Research performed a source inspection of the assemblies and was
very pleased. They have subsequently requested that all of this type of product for future
build runs be soldered using a no-clean process. A picture of the assembled circuit board
can be seen in Figure 33.
71
Note Hidden Capacitors Under Integrated Circuits
FIGURE 33: Assembled Real World Test Circuit Board
Overall, the confinning experiment results for this real world test on an actual customers
product perfonned very well.
8.9 IMPLEMENTATION CONCLUSIONS AND RECOMMENDATIONS
The results found for the standard pitch SMT components was so robust that it did
not matter what the parameter settings were set to within the range specified. Further
analysis for the ultra-fine pitch condition would have to be completed before any further
comments could be made regarding that area. The fine-pitch results showed that there
was a large bias in the experiment resulting in the majority of the characteristic effect
being accounted for by the stencil thickness. This could be accounted for by the
thickness levels being too wide or the other factors levels not being wide enough for the
fine-pitch case. With the large number of other possible factors contributing and their
potential interplay, further analysis would have to occur to be certain. Despite the
somewhat confusing results found for this case, setting the parameter levels to their
average, choosing the 8 mil thick stencil, and selecting paste J worked well for a
72
combination of standard and fine pitch components for the confirming experiment. It is
recommended that these settings be used as a starting point for all boards and the settings
be tweaked as necessary. If solder volumes are the largest difficulty encountered, the
stencil design would most likely be the suspect as supported by the fine-pitch ANOVA
analysis.
8.10 FUTURE IMPLEMENTATION AT TRISTAR TECHNOLOGIES
While the initial implementation attempt at Tristar went well, there are many
other areas that need to be addressed. First and foremost, the no-clean SMT process that
was specified needs to be attempted and successfully run for additional customer
products. Issues such as the solderability of the components, handling of the PCB's and
components, and inspection were all closely monitored and controlled for the test case.
Further control mechanisms and procedures need to be implemented so that these areas
are controlled for all assembly jobs that are to be utilized with no-clean. Education of
personnel as to these issues and how they affect them must occur also. Finally, the areas
of assembly beyond the scope of the SMT process, such as through hole wave soldering,
hand assembly, and SMT and through hole touch up must also be addressed.
The groundwork for further implementation of no-clean at Tristar has been laid
through this study. It is up to Tristars managers, assembly personnel, and customers as to
where the path will go from here.
73
CHAPTER 9: FINANCIAL IMPLICATIONS OF CONVERSION
9.1 INTRODUCTION
One of the main reasons for converting to a no-clean operation is to benefit from
the economic advantages of implementing such a process. Potential savings stem from
the reduction, or elimination of processing time, equipment, and material costs.
Obviously however, every manufacturing setting is different, so financial calculations
should be approached as such.
9.2 FINANCIAL IMPACTS AT TRISTAR TECHNOLOGIES
For the specific situation of conversion at Tristar, it was decided that the financial
factors relating to the cleaner usage and the soldering materials would be analyzed. More
specifically, each financial expenditure under each topic would be separated, calculated,
and recombined so that an over-all cost savings relative to the desired percent conversion
could be discerned.
9.3 AQUEOUS CLEANER RELATED COSTS
For their current cleaning process, Tristar employs a Trieber TRM-SMD Series
400 in-line aqueous cleaning system with a complementary closed loop water recycling
system. The in-line cleaning portion of the system uses a serial cascading cleaning then
drying scheme. The final washing zone utilizes the cleanest water which is then pumped
to the previous wash area, and repeated. A picture ofthe Trieber in-line system can be
seen in Figure 34.
74
FIGURE 34: Trieber TRM-SMD Series 400 In-line Cleaner
The recycled water which is fed in to the in-line portion of the system originates from the
recycling system. The recycling system can be seen in Figure 35.
FIGURE 35: Trieber Water Recycling System
75
The separate cost factors relating to cleaner system usage at Tristar that were
broken down can be seen in Table 22.
TABLE 22: Tristar Cleaner System Related Cost Breakdown
Electricity Consumption
Water Consumption
Processing Labor During Cleaning Operations
Filter Replacement
Miscellaneous Maintenance Materials
Maintenance Labor
Floor Space
Equipment Purchase Price
In order to facilitate a final total, the cost of each of these factors must be calculated
separately using the same units. The standardized units chosen for this analysis were
relative cost per work hour.
9.3.1 ELECTRICITY CONSUMPTION
The electricity consumption of the aqueous cleaner is somewhat variant and
dependent upon many factors. The major factors affecting this include how often the
washer is actually utilized for assembly cleaning, and how often the internal heaters are
used to keep the water at set temperatures. Due to the high mix of products and volumes
at Tristar, the throughput and work in progress vary widely and thus washer usage
consequently varies. In order to make any calculations in this situation, certain
assumptions were made. These assumptions included:
• Cleaner usage is approximately 25% of manufacturing time during working day
• Electricity cost is $0.10 per kWh
• Internal heaters are utilized for 50% of cleaner operation time
76
Once these assumptions are made, the calculation of electricity cost per hour of cleaner
system usage becomes straight forward and can be seen in Table 23 below.
TABLE 23: Aqueous Cleaner Electricity Consumption Per Hour
Instantaneous power consumption
Actual reading taken with watt
1.6kW
with heaters running:
meter from cleaner
Instantaneous power consumption
Actual reading taken with watt
without heaters:
meter from cleaner
Average instantaneous power
(l.6kW + 0.5kW) /2
1.05 kW
Power consumption per hour
1.05kWh * 25%
0.2625 kWh
Total electricity cost per hour of
0.2625kWh * $O.l/kWh
0.5kW
consumption:
manufacturing:
Therefore, the electricity cost for operating the cleaning system is $0.026 per given work
hour, based upon the assumptions made.
9.3.2 WATER CONSUMPTION
Being a closed loop system, the cleaning system should theoretically not require
the influx of any additional water volume because of recycling. Unfortunately however,
water is lost by evaporation through vents, and entrance and exit areas which are open.
This leads to the inevitable fact that water is input to the system as needed to replace the
evaporated water. Fortunately for this calculation, there is a built in inflow water meter
on the inflow water pipe. Assumptions made for calculation included:
•
•
•
•
The built in volume flow meter is calibrated correctly and yielded correct readings
There are 52 work weeks in a year with 5 work days per week
There are two 8 hour shifts utilized per work day
Deionized water in this industrial setting costs $0.05 per gallon
77
The calculation of the water consumption cost per hour based on these assumption can be
seen below in Table 24.
TABLE 24: Water Consumption Cost Calculation
Gallons consumed between
Actual reading taken from influx
8/1/94 (installation date) and
volume flow meter
9821 Gallons
7/1/96
Work hours between readings
* 5 Days / Week
500 Days * 16 Hours / Day
8000 Work Hours
Gallons consumed per work hour
9821 Gallons / 8000 Hours
1.23 Gallons /
Work days between readings
100 Weeks
500 Work Days
Work Hour
Water cost per work hour
1.23 Gallons
* $0.05 / Gallon
Therefore, the water consumption cost per given work hour is $0.06 per hour based upon
the assumptions made.
9.3.3 PROCESSING LABOR DURING CLEANING
A portion of an assemblers time is required operate the cleaner for cleaning
operations. This time includes placing and removing the assembly in to the washer and
monitoring the system during cleaning. To perform this calculation, the assumptions
included:
• Cleaner is utilized 25% of manufacturing time per hour
• The average assembler labor rate is $8.00 per hour
The calculation for processing labor cost required for cleaner operation is therefore:
0.25 Hours Usage / Work Hour
* $8.00 Labor / Work Hour = $2.00 Labor / Work Hour
78
9.3.4 FILTER REPLACEMENT
Although there is no steady stream of waste effluent from this type of recycling
system, the filters do periodically wear out and require replacement. There are two types
of filters used in series for the recycling system; ion exchange filters to remove ionized
materials, and mechanical filters to remove particulate matter. The time between .
replacement of the filters is dependent upon the actual amount the cleaner is used and the
volume of the ionic materials cleaned off of assemblies. Since this varies considerably
for Tristars manufacturing setting, the assumption that filter replacement intervals occur
every six months was made. This is based upon the empirical intervals for replacement
seen at Tristar, which have been close to that value. Other assumptions included:
• The washer is utilized 25% of a given work hour
• There are 52 work weeks in a year with 5 work days per week
• There are two 8 hour shifts utilized per work day
The calculation of the cost of filter replacement can be seen in Table 25.
TABLE 25: Filter Replacement Cost Breakdown Per Work Hour Used
Cost per filter change
Quote from manufacturer
$2000.00
Work days in six month
26 Weeks * 5 Work Days /
130 Work Days
period
Week
Hours utilized in six month
130 Days * 16 Work Hours
period
/Day
Cost of filter change per
$2000.00/2080 Work
work hour used
Hours
2080 Work Hours
The filter replacement cost relative to each work hour used is therefore $0.96.
79
9.3.5 MISCELLANEOUS MAINTENANCE MATERIALS
Due to wear and tear, other materials besides filters require replacement
periodically. These include pumps, gaskets, hoses, and other similar components. Based
upon maintenance records for the period of the previous year, the maintenance materials
required to maintain operation can be estimated. The assumptions made for this
calculation include:
• The materials required for the previous year are representative of costs every year
• There are 52 work weeks per year, 5 work days per week, and 16 work hours per day
The calculation of the material cost per work hour can be seen in Table 26.
TABLE 26: Miscellaneous Cleaner Material Maintenance Cost
Total misc. material cost for
Based on receipts
$473.00
* 5 Work
Days/Week * 16 Work
4160 Work Hours
fiscal year '95
Work hours per year
52 Weeks
Hours/day
Misc. material cost per
$473.00/4160 Hours
work hour
Therefore, the total miscellaneous material cost for maintaining the cleaner is $0.11 per
work hour.
9.3.6 MAINTENANCE LABOR COSTS
To perform maintenance on the cleaning system, maintenance personnel labor is
utilized. To accurately determine the percentage oftime maintenance personnel work on
the cleaning system is very difficult to determine. The maintenance logs at Tristar do not
80
reflect the amount of time spent, but rather only the dates and specific problem fixed.
Conversations with maintenance personnel led to the guesstimate of total percentage of
work time spent on cleaner to be 2.5% of each work day. The assumptions for this
calculation included:
• The percentage of a given work period that is spent maintaining the cleaner is 2.5%
• The average maintenance labor cost is $12.00 per hour
The calculation for total maintenance labor cost per hour is therefore:
$12.00 Labor / Hour
* 0.025 Hours Labor / Work Hour = $0.30 / Work Hour
9.3.7 FLOOR SPACE
Although the building and floor space is owned by Tristar, it constitutes space
which could be utilized alternately. The value of the floor space must therefore be
considered in the calculations. Assumptions made for this calculation include:
• Total floor space occupied by in-line and recycling system with surrounding access
areas is approximately 1000 square feet
• Floor space cost per year = $1 O.OO/square foot
The calculation of the floor space cost per hour is therefore:
(1000 Square Feet Total
* $lO.OO/Square foot per Year) / (365 Days/Year * 24
Hours/Day) = $1.14/Hour
It should be noted that this floor space cost is entirely independent of the degree to which
the cleaner is utilized. The floor space cost is present as long as the washer is occupying
space on the floor, and can not be eliminated unless 100% conversion to a no-clean
process occurs.
9.3.8 CLEANER EQUIPMENT CAPITAL COST
The original cost that was spent to purchase the cleaning system with all its
options was $76,010.00. Because this is a one time cost, it can not be equated to relative
cost per hour. Savings from the purchase price of the cleaner equipment can only be
81
realized if 100% conversion to a no-clean process is implemented, and thy cleaner can be
eliminated.
9.3.9 AQUEOUS CLEANER TOTAL COSTS
TABLE 27: Total Cleaner Operation Cost Per Work Hour
Electricity Consumption
$0.026
Water Consumption
$0.06
Processing Labor During Cleaning Operations
$2.00
Filter Replacement
$0.96
Miscellaneous Maintenance Materials
$0.11
Maintenance Labor
$0.30
Floor Space
$1.14
Capital Equipment Purchase Price
Not Applicable
TOTAL
Therefore, the total cost to operate the cleaning system per work hour is $4.60, based
upon the assumptions made. Essentially what this number means is that for every work
hour, all the associated costs for running and maintaining the cleaner total $4.60 per work
hour for Tristars current process.
9.3.10 CLEANER OPERATION SAVINGS FROM PERCENTAGE
CONVERSION TO NO-CLEAN
If a certain percentage of product types are switched to a no-clean process, the
cleaner system would not be used for those products. This reduced usage would therefore
result in a corresponding decrease in cleaner operation cost per work hour. To determine
this operational cost savings, the factors which are independent of usage must be
subtracted from the total. In this case, the cost factors which have no dependence
82
upon the usage of the machine include the floor space and the capital investment. Since
the capital cost is not applicable to this calculation, only the floor space cost is a constant
and is subtracted from the total. Therefore the equation to determine the cleaner
operation savings for a given fraction of conversion is:
Y - (((1-X)
* Z) + C) = Cleaner Operation Savings Cost For Given Percent
No-clean Conversion Per Work Hour
Where:
X = Fraction of production converted to no-clean
Y = Total cleaner operational cost for current cleaning process<fl
Z = Variable cleaner operational cost for current cleaning process<fl
C = Volume independent (constant) cleaner operational cost<fl
cp: Normalized to units per work hour
Example; If 25% of production was converted to no-clean, then the cleaner operational
cost savings would be:
4.60 -(((1-0.25)*(3.46»+1.14) = $0.86 Per Work Hour
Cleaner operational cost savings for any fraction of no-clean conversion would be
calculated the same way.
9.4 NO-CLEAN SOLDERING MATERIAL CONVERSION COSTS
Unfortunately, no-clean solder pastes tend to be slightly more expensive than their
cleaned counterparts. Instead of cost savings in this case, some additional costs are
incurred when no-clean solder paste is utilized. To quantify this, the throughput and the
amount of solder paste used for each product must be known. For a contract
manufacturing situation such as Tristar, these values vary widely and representative
assumptions must be made.
To establish some basis for quantities of paste used per board, the weights of paste
deposited on to each of ten different product types were weighed and averaged. The
average weight was found to be 15 grams. This value was used as representative.
Looking at the shipping backlog for a one month period, a representative quantity of
83
SMT boards assembled per day of approximately ten was found. Dividing 10 boards by
an assumed 16 hour work day yields an average of 0.625 SMT assemblies produced per
work hour. This means that a total of 9.375 grams of paste are used per work hour
(0.625 units times 15 grams per unit). To be converted to cost liability per work hour, the
difference in price between no-clean current aqueous clean pastes had to be found.. Based
upon standard purchase quantities of Tristar, the costs of no-clean (PASTE C) and
aqueous (TRISTAR CURRENT) pastes are $0.229/gram25 and $0.211gram respectively.
Therefore, the cost liability of using no-clean solder paste based upon these assumptions
is:
Solder Paste Cost Liability of Conversion =
-(X)«price of no-clean paste/gram - Price of current paste/gram)*Grams used per Hour)
Where:
X = Fraction of production converted to no-clean
The negative sign is indicative of the liability
Example; If 25% of SMT production were converted to no-clean solder paste, then the
increased cost of no-clean solder paste would result in a cost liability per work hour of:
-(0.25)(($0.229-%0.21)*9.375) = -$0.045 Per Work Hour
Although extreme assumptions were made to determine this value, any fraction of
conversion could be calculated the same way.
9.5 TOTAL COST SAVINGS OR LIABILITY FOR NO-CLEAN CONVERSION
To find the total savings or liability between the solder paste and the cleaner
operational costs, the two numbers are simply combined. The calculation is simply:
Total Cost Savings per Work Hour =
(Cost of cleaner operation per work hour) + (Cost of solder paste conversion per work hour)
25
Purchase quote from Dyess Supply Company, Fort Worth Texas, 7/28/96.
84
For the example of25% conversion:
Total Cost Savings per Work Hour =
($0.86 + (-$0.045)) = $0.815 per Work Hour
It should be noted that if 100% conversion to a no-clean process were accomplished, then
the constant costs such as capital and floor space costs would be considered savings and
not liabilities.
This technique presented made many assumptions, which mayor may not be
realistic. In such a variable situation as contract manufacturing, without considerable
long term data to substantiate values for these parameters, these types of assumptions
must be made. Although this technique only takes cleaner savings costs and solder paste
liability costs in to consideration, it should be noted that there are other less easily
quantified considerations of conversion which must be taken into account.
9.6 FINANCIAL IMPLICATIONS OF CONVERSION CONCLUSIONS
The impetus for conversion to a no-clean process may be driven by many factors,
as previously mentioned. If cost is the main driver, it must be understood that in any
manufacturing setting, the up front cost will likely be high but can potentially open the
door for a more cost effective process after conversion. Because every manufacturing
setting is different, the economic analysis for each should be approached as such. There
is no right or wrong way to perform a financial analysis for conversion. The procedure
presented in this study is merely one way of approaching quantifying conversion. It only
considers the cost savings and liabilities resulting from less cleaner operation when a
fraction of production utilizes no-clean and the change in solder paste costs for a no-clean
process.
As is usually true for any new process change, there are many difficult to quantify
areas where costs may be saved or incurred. Some of these areas include the potential
throughput increases from eliminating the cleaning step from processing and the up front
engineering time spent qualifying and changing to the new process. For the purposes of
this study, these areas were not addressed. For a more complete listing of savings or
liability concern areas, see the section in Chapter 4 on no-clean advantages and concerns.
85
CHAPTER 10: OVERALL CONCLUSIONS AND
RECOMMENDATIONS
10.1 OVERALL CONCLUSIONS
No-clean soldering is a viable alternative to cleaning assembled PCB's.
Utilization of no-clean solder pastes will allow many benefits to be realized. Some of
these benefits include the elimination of all previous cleaning waste effluents, the capital
investment along with other associated costs of cleaning equipment, increased
throughput, and many others. To obtain satisfactory results however, the many concerns
of no-clean soldering must be addressed such as firm up front commitment, proper
testing, qualification and education of associated personnel. For many cases, existing
manufacturing equipment can be utilized successfully with no-clean solder pastes. Some
processing changes will undoubtedly have to be made though, to ensure equal or better
quality than current processes. Some of these areas include tighter incoming material
controls, handling practices, and more frequent audits. With proper care and time, noclean soldering can be successfully implemented in to both high and low quantity
assembly environments with small or large product mixes.
10.2 RECOMMENDATIONS FOR FUTURE RESEARCH
While many qualification tests were performed for this study, this implementation
strategy is merely one approach to implementing no-clean soldering for SMT. Some
areas of research that were over looked during this examination include:
-Comparing current and future stencil manufacturing/design technologies and the
performance of no-clean pastes with them
-The effect on no-clean residue composition resulting from reflow profile changes
- The effect of soldering performance when utilizing an inert reflow environment such as
nitrogen
-No-clean wave soldering and hand soldering
-Conformal coating adhesion and testing issues
-Solderless no-clean (conductive polymer) materials
86
-Other evaluation tests that were not completed for this study
As new innovative soldering materials and processes push the boundaries of electronics
assembly, there will always be countless areas of research that may be accomplished.
87
APPENDIX A
TABLE A-1: Solder Balling Test Results
u(HOURS)~
o
2
3
4
Average
~~~~~i"+fAiffi8·1i.p~ASSTnEfi
J>
:AS
.... TIME ....
7
4
8
5
11
10
3
3
5
7
8.2
9.2
7
3.8
16.2*
30*
14.2*
* Designates Unacceptable Solder Balling Levels
8.6
4.6
TABLE A-2: Interaction Effect on Average For Squeegee Pressure and Speed
Average Analysis: Squeegee Pressure & Speed Interaction
Stencil Printer Analysis
24.5 -,--
---,
__~==4
24.1
±=========:::::===~=-~
23.7
+ - - - - -__- - - - - - - -__-.....;..;.."::::;:.0...-:::;'--1
1-..-81
I
23.3 f------"'-.....;..;..-=-=__=..:..-----.....;..;..--------j
22.9
-=-
. . .;. ;. __- __
22.5 -!--....o-_ _....;.,.,--...._....o-
!-+-82
--~.....;..;..----'--I
--....
~
A2
A1
FIGURE A-1: Interaction Effect on Average For Squeegee Pressure and Speed
TABLE A-3: Interaction Effect on Average For Squeegee Pressure and Material
I-A
4.4
Average Analysis: Squeegee Pressure & Material
Stencil Printer Analysis
25.5,.25
--,
-!--
-:::;;;>""..,.
-------=....-:::::.---------I
24.5 -!24 -!-
23.5
23
!
-:::;;;;-"~
_'___'___
_'___'__
-+-D1
_ _ 02
----I
t:~:~=====~;;:::====:J
22.5
-1.-
-----'
A2
A1
FIGURE A-2: Interaction Effect on Average For Squeegee Pressure and Material
TABLE A-4: Interaction Effect on Variability For Squeegee Pressure and Speed
VariabilUy Analysis: Squeegee Pressure & Speed
Stencil Printer Analysis
55 ,...--
----
---,
50
+-------------------------__--'--"':;;;;oo~---1
45
-!--
~-=-------------;
-+-81
_ _ 82
40
35
30
+-__-:::;;;>""~
-------'---_,_-'---'--
~-----1
+::==================J
---l
J...;..
A2
A1
FIGURE A-3: Interaction Effect on Variability For Squeegee Pressure and Speed
2-A
TABLE A-5: Interaction Effect on Variability For Squeegee Pressure and Material
~A2
33.79
47.03
39.72
42.62
Variability Analysis: Squeegee Pressure & Material
Stencil Printer Analysis
50
-r--------------------------,
48
46
i
44
42L__--~~------~
40
-+-01
___ 02
38
36
34
32
30 -'--
--1
A2
A1
FIGURE A-4: Interaction Effect on Variability For Squeegee Pressure and Material
3-A
APPENDIXB
IPC Industry Standard Test Methods
IPC-TM-650-2.4.35: Solder Paste-Slump Test
The Institute for InterconMCtlng end Packaglng Electronic ClrcIIIts
l3aO N. l.lncoln Avenue· l.lncolnwood, Il. 60646-1705
Number
2.4.35
Subject
Solder ....~ Teat
I
Date
IPC-TM-650
TEST METHODS MANUAL
1.0 SCope
This procedure determines vertical and horizon-
tal slump for solder pastes.
2.0 Applicable Dcacumenta
Revision
1/115
Originating Task Group
Solder P..to Talk Group (5-22b)
One test specimen shalt be marked as specimen #1
and one specimen as #2 and procassed in accordanCe with
paragraphs 5.2.1 and 5.2.2.
5.1.2
None
3.0 Te.t Specimen A standard specimen shall be prepared using a clean frosted glass microscope slide measuring
7.6 cm x 2.5 em, minimum 1 mm \hick. M equivalent alumina
or glass epoxy substrate may be used.
4.0 EqulpmentlAppanltua
5.2.1 The specimens shall be stored for 10 to 20 minutes at
25 +/-S°C and 50% relative humidity +1-10% and specimen
#1 examined for siump.
5.2.2 Specimen #2 from 5.2.1 shall be heated to 150
+1-lO°C for 10 to 15 minutes, cooled to ambient and examined for slump.
Stencils
Ipc·A·21, Ipc·A-20
5.3 EYIIluation Enter data in Table 1 and/or Table 2 by
entering spacings which have bridged with a suitable check
Steel Squeegee (razor blade)
Oven
mark.
Microscope
5.0 ProcecIure
Specimen preparation using appropriate stencil pat·
tem IPC-A-21 or Ipc·A-20. (FlQures 1 & 2) Deposit solder
paste patterns on 2 substrates for each stencil pattern. The
printed pattern shalt be LIllform in thickness with no solder
particles separated from the pads. The vendor and user
should use the same printing method.
5.1.1
Table 1
Tabla 2
Stencil IPC-A·21 (0.2 mm Thlck)
Stencil IPC-A-20 (0.1 mm ThiCk)
PIId slZlI 0.63 x 2.03 mm
Spacing
mm
PIId slZlI 0.33 x 2.03 mm
Spacing
Hor.
Vert.
mm
Pad Ilu 0.33 x 2.03 mm
Spaclng
Her.
Vert.
mm
0.79
0.45
0.45
PIId ale 0.2 x 2.03 mm
Spacing
Hor.
Vert.
mm
0.30
0.71
0.40
0.40
0.25
0.63
0.35
0.35
0.20
0.56
0.30
0.30
0.175
0.48
0.25
0.25
0.15
0.41
0.20
0.20
0.125
0.33
0.15
0.15
0.10
O.lG
0.10
0.075
0.08
0.08
I-B
Hor.
Vert.
IPC-TM-650-2.4.35: Solder Paste-Slump Test Continued
IPC-TM-Gi50
Dale
SuOject
<f;umber
SOlder _ u m p Tnt
2.4.35
ltv5
Revision
Spacing -- AI
'pacing
...
..
...
.11
"
11
...
...
..
".......
Spacing
@§~DIDIDIDIDID~D
Pad'i,e,
o.36mm_~
Vertical
0.63 x 2.03mm
~::::~.- Rows -+C::J -14 identical.pads per row
~:::::=
c::J -Same spacings each row
c::J
O.30"",,--"c::::::::J
o.251M'1--+-§5
Spacing
~:~
=:::::r:::=:::l
c::J
c::J
g:~:::=:::
c::J
c::J
c::J
c::J
c::J
Pad Size:
0.33 x 2.03 mm
-18 identical pads per row
-Same spacings each row
Flgu.. 1 Slump Itlt .,_11, IPe-A·2l
Pad Size: 0.20 x 2.03 mm '""""------,
-16 identical pads per row
-Same spacings each row
Spacing
o.oo""'_~
o.10rrm-.~
~~~~~ [ill~~~~~~~~
O.35mm--' c::=:=J
~
O'o4O...-------c::=:=J
o.o4Omm---.~
O.3$mm--.
.....
o.3Onwn--.§5
o.25_--..c::=:=J
o..2Omm--...c::=:=J
0.15
c::=:=J
nwn-.
Spacing
~::::::::::
o.m_
~E~~~":,.
O.«imm--...
Vertical
Rows
Spacing
~:=::~::W
~4-0.z0_
~::::::
c=:=:::J:::::::
~,,--o.l761m1
4 - - Q.IJImt
~E~-E:
~~~,~~~~~~~~~
Spacing
-18 identical pads per row
-Same spacings each row
Figur. 2 Slump _
ItInClI, IPC-A·20
2-B
IPC-TM-650-2.4.45: Solder Paste- Wetting Test
TIle Instltulll tor Inll8l'COllnec:tlng and Packaglng Elec1ronic Clrcuitl
7380 No Uncoln Avenue· UncoInwood, IL &064&-1705
Number
2.4..44
Subject
Solder Paate-Wetllng Teat
Date
IPC·TM·650
TEST METHODS MANUAL
1.0 SCope Determine the ability of a solder paste to wet an
oxidized copper surface and to qualitatively examine the
amount of spatter of the solder peste during re1low.
2.0 Applicable Documents None
IPC-TM-eSO Test Methods Manual
Originating Task Group
Solder Paste TaUt Group (l5-22b)
5.2.3 After reflow, the residual flux shall be removed with a
SlitabIe solvent.
5.3 Evaluation When examined visually at , OX, the solder
shall uniformly wet the copper and there should be no evidence of dewetting or non-wetting of the copper and there
shall be no solder spatter around the printed dots.
2.4.43 Solder Paste-Solder Sal Test
3.0 Teat SpecImen
7.6 em x 2.5 em x 0.8 mm specimen of 1 ounce oxygen-free
high conductivity (OFHC) copper.
4.0
RENislon
1/M
IiquI~slApparatus
Flat hot plate
Specimen tongs
Beaker 400 cc
Magnifying glass with '0 times magni1ication
Uquid copper cleaner
Deionized water .
Isopropyl alcohol
Solvent for residual flux removal
4.1 Stencil 76 mm x 25 mm x 0.2 mm provided with at least
3 round holes or 6.5 mm diameter aperature with a minimum
between centers of 10 mm.
5.0 Procedure
5.1 Preparation
5.1.1 The specimen shall be cleaned with a liquid copper
deaner, washed thoroughly with water, rinsed with isopropyl
alcohol, dried and then placed in boiing deionized water for
10 minutes and air dried
5.2 Teat
5.2.1 Place stencil on test specimen and print solder peste
test pattern.
5.2.2 Reflow using the procedure outlined in paragraph
5.2.3.2 of Ipc·TM-650, Test Method 2.4.43.
3-B
IPC-TM-650-2.6.15: Corrosion, Flux
The Institute for Interconnecting and Packaging Electronic Clrcutts
7380 N. Lincoln Avenue· Lincolnwood, IL 6064&-1705
Number
2.8.15
Subject
CoITOIIlon, Flux
Date
1/95
IPC·TM·650
TEST METHODS MANUAL
Revision
B
Originating Task Group
Flux SpecIfIcations Tuk Group (5-22d)
1.0 Scope This test method is designed to determine the
corrosive properties of lIux residues under extreme environmental conditions. A pellet of solder is melted in contact with
the test lIux on a sheet metal test piece. The solder is then
exposed to prescribed conditions of humidity and the resulting corrosion. if any. is assessed visually.
5.1.1.1 Ammonium persulphate (25% mlv in 0.5% vlv sulfuric acid). Dissolve 250 9 of ammonium persulphate in water
and add eautiousIy 5 ml of 5% sulphuric acid (relative density
1.84). Mix. cool. dilute to 1 liter and mix. This solution should
be freshly prepared.
5.1.1..2 Sulfuric acid (5% vlv). To 400 ml of water cautiously
add 50 ml of sulfuric acid (relative density 1.84). Mix. cool.
dilute to 1 liter and mix.
2.0 Applicable Documents
IPC-TM-6S0 Test Methods Manual
2.3.34 Solids Content. Aux
British Standard lnatltute BS5825 Spec:ification of Purchasing Requirements and Methods of Test for Auxes for Soft Soldering
3.0 Tnt Speclmen At least 0.035 9 of lIux solids. 1 9 solder peste. 1 9 wire. or 1 9 preform with an eqUvaJent amount
of solids. Flux solids are defined as the residue from the s0l-
ids content. fluxes test described in IPC-TM-650. Test
Method 2.3.34. All solvent must have been evaporated from
the specimen in a chemical fume hood.
5.1..2.1 Cut a 0.50 ±a.os mm thick piece of 99% pure c0pper 51 mm x 51 mm for each test.
5.1..2..2 Form a circUar depression in the center of each test
panel 3.2 mm deep by forcing a 19 mm steel ball into a 25.4
mm hole to form a cup.
5.1..2.3 Bend one comer of each test panel up, to facilitate
subsequent handling with tongs.
4.0 Apparatus and ReqentlI
5.1.3.1 Immediately before performing test. pretreat as follows using clean tongs for handling.
1. Solder pot
2. Humidity chamber capable of achieving 50 ±2"C and 95
2% relative humidity.
3.
~r
circulating drying
±
oven
4. Microscope having 20X minimum
5. Chemicals: All chemicals must be reagent grade and water
must be distilled or demineralized:
a. Ammonium persulphate
b. Sulfuric acid. % vlv
c. Degreasing agent. Acetone. Toluene. or Petroleum
ether.
A. Degrease with a suitable neutral organic solvent such as
acetone. toluene. or petroleum ether.
B. Immerse in 5% sulfuric acid (by volume) at 65 ±SoC for 1
minute to remove the tarnish film.
C. Immerse in a solution of 25% mIv ammonium persulphate
(0.5% vlv sulfuric acid) at 23 ±2°C for 1 minute to etch the
surlace lXliformly.
D. Wash in ruming tap water for
E. Immerse in 5% sulfuric acid
a maximum of 5 seconds.
(by volume) at 23 ±2°C for 1
minute.
6. Analytical balance capable of weighing 0.001 9
F. Wash for 5 seconds in running tap water. then rinse thoroughly in demineralized water.
5.0 Procedure.
G. Rinse with acetone.
H. AJbw to dry in clean air.
5.1 Preparation
5.1.1 Chemical.
5.1.3..2 Use the test piece as soon as possible or store up
to 1 hour in a closed container.
4-B
IPC-TM-650-2.6.15: Corrosion, Flux Continued
IPC-TM-650
Number
2.8.15
Date
1/i5
Subject
CoITOSion, Flux
Revision
B
5.1.4 Preparation of Tnt Solder
5.1.4.1 Weigh 1.00 gram iQ.05 gram specimen of solder for
each test and place in center of depression of each test panel.
5.1.4..2 Degrease solder specimen with a suitable neutral
organic solvent such as acetone. toluene. or petroleum ether.
5.1.4..3 Solder may be in tI1e form of pellets or by forming
tight spirals of solder wire.
5.2 Tnt
5..2.1 Heat solder pot so that solder bath stabilizes at 235 ±
S·C.
5..2.2 Uquld Flux Place 0.035 g of flux solids into the
depression in the test panel. Add solder sample.
5.2.& Expose specimens to the above environment for 240
hours (10 days). M and H may be tested in the cleaned. as
well as uncleaned. condition.
5.3 Evaluation Carefuly examine specimens prior to placing them in the environmental chamber. Note any discolora-
tion.
5.3.1 After the appropriate exposure period. remove test
specimens from humidity chamber. examine at 20X magnification and compare with observations noted in paragraph
5.2.5.
5.3.2 Corrosion is described as follows:
A. ExcrescenCes at the intertaces of tI1e flux residue and capper boundary, or tI1e residues or discontinuities in tI1e residues.
8. Discrete wi'1ite or colored spots in the flux residues.
5.2.2.1 Solder Pute, CoredaWlre or CorecI-Prefonn
Place 1 g of solder paste, flux-cored wire or cored-preform
into the depression in tI1e test p8ne1.
5.2..3 Using tongs. lower each test panel onto tI1e surface of
tI1e molten solder.
5..3..3 />on initial change of color which may develop IM1en the
test panel is heated during soldering is disregarded. but subsequent development of green-blue discoloration with observation of pitting of the copper panel is regarded as corrosion.
8.0 No. .
5..2.4 Allow the test panel to remain in contact until soIcler
. specimen in tI1e depression of tI1e test panel melts. Maintain
this position for 5 ±1 seconds.
5..2.5 carefully examine test specimen at 20X magni1lcation
for Slbsequent comparison after humidity exposure. Record
observations. especially any ciscoloration.
8.1 Definition of Corrosion For purposes of this test
method. the following de1inition 01 corrosion shall prevail. "A
chemical reaction between the copper, the solder, and the
constituents of tI1e flux residues, wi1ich occurs after soldering
and during exposure to tI1e above environmental conditions."
5..2.8 Preheat test panel to 40 ±1"C for 30 ±2 minutes.
8.2 Color photos before and after tI1e test are valuable tools
in identifying corrosion. (see 5.2.5.)
5.2.7 Preset humidity chamber to 40 ±1·C and 93 ±2%
relative humidity.
8.3 Sa~ Observe all appropriate precautions on MSDS
for chemicals involved in this test method.
5.2.8 Suspend each test specimen vertically (and separately) in tI1e hUmidity chamber.
5-B
IPC-TM-650-2.4.43: Solder Paste- Solder Ball Test
The Institute tor Interconnecting and Packaging Electronic Circuits
7380 N. Uncotn Avenue' Uncotnwood, IL 60646-1705
Number
2.4.43
Subject
Solder P.ste-Solder B.II Test
Date
Revision
1/05
IPC·TM·650
TEST METHODS MANUAL
Originating Task Group
,
Solder Paste T••k Group (5-22b)
1.0 Scope This test is carried out to determine the reflow
properties of the solder peste. The ability of the preaJloyed
solder perticles in the peste to reflow into a sphere on a nonwettable substrate is determined under defined test condi-
5.1.3 Condition the peste to uniform temperature of 25" C
tions.
+/-2°C.
2.0 Applicabl. Documents
None
3.0 Test Specimen
Frosted glass microscope slide. alumina substrate or glass/epoxy printed circuit board with a
thickness of 0.60 to 0.60 mm and a minimum length and
width dimension of 76 mm and 25 mm, respectively.
5.1.2 Homogenize the solder peste by hand stirring with a
spatula.
5.1.4 Prepare two test specimens with either/or both stencils listed above (4.1.1 and 4.1.2). The solder peste should be
squeeged with the spatula to fill and level each hole.
5.2 T••t
5.2.1 T••t Conditions
4.0 equipment/Apparatus
4.1
Metal Stencils
5.2.1.1 Test one specimen within 15 +/-5 minutes after
placement of solder peste on test coupon.
4.1.1' Stencil for Type 1-4 Stencil 76 mm x 25 mm x 0,2
mm provided with at least 3 round holes of 6.5 mm diameter
5.2.1.2 Test the second specimen 4 hours +/-15 minutes
after placement of solder peste on test coupon. Storage for 4
aperttAs with a minimum distance between centers of 10
mm.
hours shall be at 25°C +/-3°C and 50 +/-10% RH.
5.2.2 CondItioning Heating equipment
4.1.2 Stencil for Type 5-8 Stencil 76 mm x 25 mm x 0.1
mm provided with at least 3 round holes of 1.5 mm diameter
5.2.2.1
Clean the surface of the solder bath with the
apertures with a minimum distance between centers of 10
mm.
scraper.
4.2 Spetula
hot plate to ensure proper control.
5.2.2.2 Remove all foreign material from the surface of the
4.3 Solder bath not less than 100 mm x 100 mm x 75 mm
5.2.3 Solder Reflow Reflow specimens by one of the fol-
deep containing solder suitable to maintain a temperature of
25°C above the liquidus temperature of the solder peste being
lowing
evaluated.
5.2.3.1 Lower the substrate. in a horizontal position with the
paste deposit on top, into the solder bath at a speed of 25 +/-
4.4 Flat hot plate
4.5 Surface temperature thermometer
4.8 Magnifying glass with a 10 to 20 times magnification.
5.0 Procedure
two methods.
2 mmlsecond until the substrate is 50% submerged. It is
importarrt that good thermal contact is achieved between the
molten solder and the substrate. k!. soon as the solder has
melted, withdraw the substrate from the solder bath maintaining it in a horizontal position. The total time on the solder bath
. shall not exceed 20 seconds.
5.2.3.2 Place the substrate on the hot plate. k!. soon as the
solder has melted. withdraw the substrate from the hot plate
maintaining a horizontai position. The reflow shall occur within
20 seconds after the specimen is placed in contact with the
5.1 Preparation
5.1.1 Set the temperature of the solder bath or hot plate at
a temperature of 25"C +/-3°C above the liquidus temperature
of the solder alloy.
hot plate.
6-B
IPC-TM-650-2.4.43: Solder Paste- Solder Ball Test Continued
IPC-TM-650
Number
2.4.43
Subject
Solder P••te-SoIder Ball Teet
Revision
5.3 Evaluation
5.3.1 Examine the reflowed specimens under 10X to 20X
magnification.
5.3.2 Solder ball size and number should be compared with
Figure 1.
5.3.3 Record the degree of reflew in comparison with Figure
1 for the 6.5 cm and 1.5 cm acceptance/reject conditions.
respectively.
7-B
Dale
1/95
IPC-TM-650-2.4.43: Solder Paste- Solder Ball Test Continued
IPC-TM-650
Number
2.4.43
Subject
Solder Paat.-8ofder Ball Teat
Date
1/95
Revision
Ul1llCC8P1abIei Clusters
Figure 1 Solder ball teat standards
8-B
IPC-TM-650-2.3.25: Resistivity of Solvent Extract
J·S'T'D-001A
Jall\lary 1995
•
Appendix F
Resistivity of Solvent Extract
F-1 This appendix defines standard resistivity of solvent
extract test methods for determining printed wiring assembly cleanliness.
verified for correct composition upOn initial use, each
shift, and according to manufacturers recommendations. The time limit may be extended when the results
of data provide a definite indication that such actions
will not adversely affect the results of the test.
-
1'-2 GENERAL REQUIREMENTS
':-2.1 RaJ_tiv;,.,
0'
~Jvent
.nract Solvent extract
resistivity shall be measured as follows:
a. Prepare a test solution of 75+0/-2 percent by volume of
reagent grade isopropyl alcohol, with the remainder
being deionized water. Pass this solution through a
mixed bed deionizer cartridge. After passing through
the cartridge the resistivity of the solution shall be
greater than or equal to 6±Q.5 megohm-centimeters
(conductivity shall be less than 0.166 microhmslcm).
b. Use a clean plastic bag, which can be sealed, which is
free of ionic contamination. It is recommended that a
heat sealable bag (e.g., Kapak@ type or equivalmtl
which contributes no ionic contamination, be used.
Another option uses a plastic "zipper locking" type
bag, that has been rinsed with a portion of test solution. Measure 1.55 milliliters of fresh test solution for
each square centimeter of assembly area (1OmVsq.in.)
into the clean plastic bag. The assembly area includes
the areas of both sides of the board plus the components.
c. Place the circuit board into the plastic bag with the test
solution and seal the bag. The bag should be sufficiently large enough for the board, and the test solution
shall be allowed to wash the surface of the board while
in the bag. Shake the bag containing the circuit board
and the test solution for approximately 10 minutes. Let
the bag sit for 5 minutes with the assembly immersed in
solution, then shake for an additional 5 minutes. Be
careful not to let the corners of the board tear the bag
while shaking.
d. Remove the circuit assembly, and measure the ionic
contamination of the test solution. The final resistivity
shall not be below 2 megohm-eenrimeters (conductivity
is 0.5 mierohmslcm>. A resistivity change from 6 to 2
megohm-eentimeters indicates a contaminant level of
1.55 micrograms of ionic contaminant per square centimeter of board surface area.
1'-2.2 SodIum chlorlcs. ..h equivalen1 Ionic contami-
b. The equipment must be validated using a known
amount of sodium chloride standard on the same
schedule as the percentage composition verification.
c. The starting, or reference, purity of the solution shall
be greater than 20 megohm-eentimeters, (conductivity
shall be less than 0.05 microhmslcentimeter) before
each sample is tested.
d. The test length shall be in accordance with manufacturers' recommendations, but in no case shall be less than
10 minutes.
e. The final conductivity is based on reference to the NaCI
standard. However, the final ionic contamination shall
be less than 1.55 micrograms per square centimeter of
board surface area.
1'-2.3 Alternate methodll Alternative equipment, with
the appropriate equivalence values, may be used to verify
cleanliness. The contractor shall develop equivalency factors for each individual system.
NOTE: There are several commercially available ionic
cleanliness testers that have shown to be equivalent to or
more efficient than resistivity of solvent extract method of
paragraph F-2.1. These alternative equipments may be
either a static method, where the test solution in continually circulated in the test cell at a fixed volume and then
measured, or a dynamic method, where the test solution is
constantly being measured as the volume flows past the
assembly. There are many variables of each system that
will affect the final resistivity, Le., heat, solution volume,
or sprays.
NOTE: Test procedures and calibration techniques for
these methods are documented in Materials Research
Report 3-78 "Review of Data Generated with Instruments
Used to Detect and Measure Ionic Contaminants on
Printed Wiring Assemblies." Application for copies of this
report should be addressed to the Commanding Officer,
Naval Air Warfare Center, Aircraft Division Indianapolis,
Code DP3070N1MS-79, 6000 E. 21st Street, Indianapolis,
IN 46219-2189.
nation tnt Sodium chloride salt equivalent ionic contamination shall be measured as follows:
a. The sodium chloride salt equivalent ionic contamination test shall use a solution of 75+0-2 percent by volume of reagent grade of isopropyl alcohol with the
remainder being deionized water. The solution shall be
9-B
IPC-TM-650-2.3 .25: Resistivity of Solvent Extract
The Instl1utll for Intllrconnectlng Ind PIckaglng Electronic CircuItS
7380 No Uncdn Avenue· Uncdnwood, IL 6064&01705
Number
2.3.25
Subject
Reslatlvlty of Solvent Extract
IPC-TM·650
TEST METHODS MANUAL
5.3 Use 100 mililiters of wash solution for each 250 mm2
{10 in.2] of the board or assembly area and direct in a fine
stream on both sides of the test specimen. The wash activity
shal\ occur for a minimum of one (1) minute. It is imperative
that the initial washings be included in the sampie to be measured for resistivity.
5.4 Measure the resistivity of the collected wash solution in
accordance with the conductivity bridge manufacturer's
instructions.
see 6.1, References
e.o
3.0 TNt Specimen My printed-wiring board or assembly
of sufficient area to provide enough the solvent to determine
its resistivity.
4.1
Miscellaneous laboratory ware such as beakers, f\.rlneIs,
Freedom of Information Act Office
Naval Avionics center
6000 East 21st Street
Indianapolis. IN 46219-2189
4.2 Conductivity Bridge, Barnstead Model PfvI-7OCB or
equivalent.
4.3 ConductMty Cell, Beckman Model Ca-.AD1-Y87-CERT
with a cell constant K • 0.10/cm or ElqlivaIent.
4.4 Wash solution composed of 75% by volume, ASC
Reagent Grade isopropyl alc0hoi, 25% by volume distilled
water with a minimum resistivity of 6.0 x 10e ohm-cm.
4,5 Unear measuring device such as calipers' capable of
measuring the area of the printed-wiring board or assembly to
the nearest 25.4 rnrn2 {1.0 in. 2].
5.0 TNt PnIoedure
5.1 Determine the surface area of the printed-wiring board
or assembly inclucfng bo1h sides of the board and an estimate
of the area of any components mounted thereon.
Suspend the
test
specimen witlin
a
(X)l'lYElNent
Notes
e.1 Reterenca This test procedure, including solution
preparation and a laboratory ware cleaning procedure, is
documented in Materials Research Report No. 3-72.
"PrInted- wiring assemblies; Detection 01 ionic contaminants
on". Application for copies of this report should be addressed
to:
storage and wash bottles. The laboratory ware may be made
of glass or poIye1hyIene but it is imperative that all apparatus
used in this procedure be scrupuiously ctean.
U
A
11/84
1.0 Scope This test method is designed for determining the
presence of ionizable surface contaminants on printed-wiring
boards and assemblies. The basis of the test is the change in
electrical resistivity of the solvent used to wash the printedwiring board or assembly. The resistivity is lowered v.t1en contaminants, such as plating salts, flux residues, etchants, or
detergents, are dissolved.
2.1
Revision
Date
sized
funnel positioned over a coIIec:tiol'I breaker.
lO-B
IPC-TM-650-2.6.3.3: Surface Insulation Resistance, Fluxes
The Institute for Interconnecting and Packaging Electronic Circuits
7380 N. Uncoln Avenue· Uncolnwood, IL 60646-1705
Number
2.8.3.3
Subject
Surface Insulation ReaI.lance, Fluxea
IPC-TM-650
TEST METHODS MANUAL
1.0 Scope This test method is to characterize fluxes by
determining the degradation of electrical insulation resistance
of rigid printed wiring board specimens after exposure to the
speci1ied flux. This test is canied out at high humidity and heat
conditions.
2.0
~lcable DoculMnta
1PC-B-24 SUrlace Insulation Resistance Test Board
IPC-A-800 Acceptability Guidelines
Date
11/14
Revision
A
A POINElr supply capable of prodUcing a standing bias
potential of 45-50 volts DC with a tolerance of ±1 0%.
4.2
4.3 A resistance meter capable of reading high resistance
(10 '2ohms) with a test voltage of 100 volts or an ammeter
capable of reading 10.10 amps in combination with tOO volts
DC power supply.
4.4 Three 2000 mI beakers
4.5 Exhaust ventilation hood.
4.8 Metal tongs.
3.1 Comb PIIttltms Use the IPC-B-24 test pattern which
consists of four comb patterns per coupon. The individual
comb, pictured in FlQure 1, has 0.4 mm lines and 0.5 mm
spacing. The test coupon shall be unpreserved bare copper
metallization.
3.2 Laminate The laminate material for this test shall be
FA-4 epoxy-glass.
4.0 Apparatus
A clean test chamber capable of programming and
recording an environment of 25 +10/-2°C to at least 85 ±2"C
and 85% ±2% relative humidity. A salt solution and desiccator
may be used to maintain humidity if a tight temperature con·
trol is maintained on the temperature of the chamber.
4.7 Soft bristle brush
or distilled water (2 megohm-em, minimum
resistivity recommended).
4.8 Deionized
4.' Drying oven capable of maintaining at least 5OOC.
8.0 Teat
5.1 Teat Condltlona A11·fluxes"";l1 be tested at 85 ±2°C.
85±2% relative humidity for 168 hours.
4.1
5.2 SpecImen Preparation There shall be 3 test coupons
for each liquid flux to be tested in the cleaned state {see Table
1, Aj. When testing Nquid fluxes which are intended to remain
in the uncleaned state, six test coupons are required. llYee
uncleaned test coupons shall be wave soldered pattern side
down (fable 1, B) and three shall be wave soldered pattern
side up (fable 1, C).
Solder paste coupons shall be re1lowed pattern side up and
either eleaned (fable 1, D) or not cleaned (fable 1, E). In addition, there shall be at least 2 unprocessed control coupons for
comparison purposes (fable 1. Fl.
Positive, permanent and non-contaminating identification of test specimens is of paramount importance. (For
example. a vibrating scribe.)
5.2.1
8.2.2 VISUllIIy inspect U'le test specimens for any obvious
defects, as described in 1PC-A-600. If ttlere is any doubt
about the overall quality of any test specimen, the test specimen shoUd be disCarded.
Figure 1 TypIc:8I"Comb Pattam" (from 1PC-B-24)
ll-B
IPC-TM-650-2.6.3.3: Surface Insulation Resistance, Fluxes Continued
IPC-TM-660
Subject
SumIce Insulation Resistance, Flux••
Number
2.6.3.3
Dale
111M
Revision
A
Table 1 Coupons for SIR Testing
sampl. Group
A
FlUXISoldec'
B
yes
C
yes
D
yes
E
F
yes
no
no
no
yes
Clean
yes
no
no
yes
Number of
Coupons
3
3
3
3
3
2
A =Pattern do'M1lclean
B = Pattern do'MVno clean
C =Pattern uplno clean
D =Solder paste/rellow/clean
E =Solder paste/rellow/no clean
F =Control (precleaned, unprocessed)
5.2.3 Clean the test coupon with deionized or distilled water
and:scrub with a soft bristle brush for a minimum of 30 seconds. Spray rinse thoroughly with deionized or distilled water.
Rinse cleaned area thoroughly with fresh 2-propanol.
AA alternative cleaning me1hod is to place the test coupon in
an ionic contamination tester containing 75% 2-propanol,
25% deionized water and process the solution until all ionics
have been removed.
During the remainder of the specimen preparation, handle test
specimens by the edges only, or usa non-contami.nating rubber gloves.
5.2.4 If boards are to be stored before treatment. place the
boards in Kapak bags or other contamination-free containers
and close bags (do not heat seaI). (KapaI< bags are available
from FISCher, VWR and other distributors).
5.3 Procedure
5.3.1 Semple Pntperatlon Flux application and soldering.
5.:1.1.1 Uquld Flux or Flux Extnct Coat the comb pattern with a thin coating of the liquid tux or 1Iux extract under
test.
5.3.1.1.1 The test ~ shall be exposed to solder by
floating the 1Iuxed comb patterns of the test specimens face
down on the solder pot at 245-260"0 for 4 ±1 seconds. Wave
solder of comb patterns face dov«'I at 245-26O"C and a c0nveyor speed with a contact time of 3± seconds. For ftuxes to
be tested in the uncleaned state, a second sat of comb patterns shall be ftuxed and bted pattern up on the solder pot
or passed pattern up over the sdder wave.
5.3.1.2 Solder Pu. Stencil print the solder paste on to
the comb pattern using a 0.2 mm thick stancil (the IPC-A-24
artwork contains the stancil design).
5.3.1.2.1 The samples shall be run through a reflew soldering process using the temperature profile recommended by
the vendor.
5.3.2.1 After exposure to flux and solder, samples to be
tested in an uncleaned state shall be evaluated as in 5.3.3
through 5.4.1.
5.3.2.2 After exposure to 1Iux and solder, samples to be
tested in the cleaned state shall be cleaned using one of the
procedures listed below. The cleaning parameters shall be
reported in the Qualification Test Report (Appendix AI.
5.3.2.2.1 The samples to be cleaned shall be cleaned with
an appropriate enWonmentaily safe solvent or aqueous cleaning medium. The usa of a commercial in-line or batch cleaner
is preferred. If 1l1is is not available, the foIIo'Ning laboratory
cleaning process shall be followed.
Three' samples shall be cleaned (within 30 minutes or less)
after soldering. For solvent or aqueous detergent cleaning,
three 2000 ml beakers each containing 1000 ml of solvent
shall be used such that one beaker SBI'\Ies as the primar;
cleaning stage and the other two are used for rinsing purposes. Each test coupon shall be agitated in each beaker for
1 minute. In the case of aqueous detergent. one beaker shall
contain the cleaning agent and the remaining beakers shall
contain deionized water for rinsing purposes. After the cleaning procedt..re is complete, samples are dried for 2 hours at
SO"C. FoIo'Ning cleaning, the specimens shall be tested as
outlined in 5.3.3 ttYOugh 5.4.1.
5.3.3 Preparetlon of Sample. for Chamber Visually
all combs and discaro any combs with bridging of
conductors. Use water v.tlite rosin to solder tefton-insulated
wires to the connection points of the specimens. Do not
atternpt to remove the llux residues. Connectors may be used
in lieu of soldering 'Nires but are not recommended. In the
event of a dispute, the samples with soldered 'Nires shall be
used as a referee.
inspect
5.3.4 Place the specimens in the enWonmental chamber in
a vertical position such that the air bN is paraHel to the direction of the board in the chamber. Set the chamber temperature at 85±2"C and humidity at 20% RH and aliow the oven
to stabilize at 1l1is temperature lor 3 hours. Then, slov.iy ramp
the humicity to 85±2% over a minim.m 15 minute period,
12-B
IPC-TM-650-2.6.3.3: Surface Insulation Resistance, Fluxes Continued
IPC-TM-650
Number
2.&.3.3
Subject
Surt.oe Inaulation Aalatllnoe, F1uxea
Date
11/94
Revision
A
Allow the specimens to come to equilibrium tor at least 1 hour
before applying the bias voltage to begin the test. If a salt
solution and dessicator are used for humidity. specimens shall
be held for 24 hours before beginning the test.
between two electrified conductors. Both of these conditions
must be eliminated for proper testing.
8..2 IPC-B-24 test board artwork and electronic data is available from IPC.
5.3.5 Connect the 45-50'1 DC voltage source to the specimen test points to apply the bias voltage to all specimens.
5.4.1 Measlxements shall be made with test specimens in
the charnbIlr under the test conditions of temperature and
humidity at 24. 96 and 168 hours. To take these measurements, the 45 • SOY DC bias voltage source must be removed
from the test spedmen and a test voltage of ·100v DC shall
be applied. (Test voltage polarity is opposite the bias polarity.)
5.5.1 Each comb pattern on el:K:h test specimen shall be
evaluated by the insulation resistance values obtained at 96
and 168 hours. If the control coupon reedngs are less than
1000 megohms, 8 new sat of test coupons shall be obtained
and the entire test repeated. The reeding at 24 hours may fall
below the required value provlded that it reeowrs by 96
hours. My reason for deleting values (scratetles, c0ndensation. bridged conductors, outlying points, atc., must be
noted).
5.5.2 All specimens shall also be examined lXlder a lOx to
SOx microscope using bacl<Iighting. If dendritic growth or corrosion is observed. it shall bEl determined If the dendrite spans
25% or more of the original spacing. This latter c:ondtion oMll
constiMe afailure. It should be determined wt1ether dendritic
growth is due to condensation from the chamber (see paragraph 6.1).
505.3 Rejection of results for more than 2 combs for 8 given
condition shaH reql.ire the test to be repeated.
&.0 Notes
8.1 If condensation OCCU'S on the test specimena in the
enWOI rnentaI chamber vHt the samples are lXlder voltage.
dendritic growth .". oca.r. This ca'l be C8U$IlCI by a lack of
sufIicient control of 1he tunlcIIIcatlon of 1he oven. Water spotting may also be obIJer\lBd in some 0YIf'l8 where 1he air 1Iow
in the chamber is from beck 10 front. In this case. water c0ndensation on the cooler oven Mldow can be blown 8l'CUld
1he 0IIl.lI'l as mIa oci'cpleI:s v.tlich deposit on test specImena
end' cause dendritic ~ if 1he spots brlcIge 1he dstanee
13-B
APPENDIXC
List of Industry Contacts
508-960-2496
Fax: 508-960-1187
Les Hymes
PWA ~anufacturing Consultants
W137N8180 River Park Dr.
~enomonee Falls, WI 53051
414-251-4222
Fax: 414-255-8128
Thomas Sommer
T~S Consulting Co.
148 Townsend Rd.
Groton, ~a. 01450
508-448-6005
Doug Peck
Advanced Electronics Interconnect Center
25 Iris Way
Haverhill, ~a. 01830
508-373-2600
Robert Laurenti
Tristar Technologies Inc
128 ~errimack St.
~ethuen, Ma. 01844
508-794-5100
Tim O'neal
ESP solders
14 Blackstone Valley Place
Lincoln, RI 02865
1-800-338-4353
Fax: 401-333-4954
Sammy Shina
University of ~assachusettsat Lowell
One Unversity Ave
Lowell, ~a. 01854
508-934-2590
Fax: 508-877-7071
Richard Cloutier
~TX Stencils
5450 Campus Drive
Canandaigua, NY 14425
716-396-6974
Fax: 716-396-6868
Greg Tashjian
Lucent Network Systems Technologies
1600 Osgood St.
No.Purrdover,~a.01845
Dr. Brett E. Hardie
Hardie Consulting Co.
7117 Woodhollow Drive #517
Austin, Tx. 78731
512-338-9884
1-C